PHASE II FINAL REPORT VOLUME VII EDP SYSTEM
Document Type:
Collection:
Document Number (FOIA) /ESDN (CREST):
CIA-RDP78-03952A000100070001-5
Release Decision:
RIPPUB
Original Classification:
S
Document Page Count:
406
Document Creation Date:
November 16, 2016
Document Release Date:
February 1, 2000
Sequence Number:
1
Case Number:
Publication Date:
March 1, 1965
Content Type:
REPORT
File:
Attachment | Size |
---|---|
CIA-RDP78-03952A000100070001-5.pdf | 14.55 MB |
Body:
Approved For eleaseflP78O38
-0
(1,
t44, wu.
__ "?rr"- ? --.'-11L:,,--- ? . , - -
??. '?-''' ',,,,:,-,;:',.' ' ,,,.io4w.....s\
2A000100070001-
,
,
$t U FINAL REPORT
, - -
Volume VII
DP SYSTEM
r".
.1'41
? :rt." t
-
March 1965
. r
-
? 4. ????
_
1 ?t
-
m _
_i -4W ?,- v-4, ,ilt.-1, -,- ,, .. - . 7,. ,A7,41'' "1-7.?1- _ ' : ;-_, _
DIRECTORATE OF 6IENCL AND :11-LCHNOLOGY
OFFICE OF COMPUTER SERVICES
f?x?cfuci.:(d:4f45r,,,?..,"'-et-L,-.
1
L41,,wngro7inr, m.,4 i
Approved For Release 2000/05/08 : CIA-RDP78-039-52A00(liocto7opoi;-
' I. It44
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
WARNING
This material contains information affecting
the National Defense of the United States
within the meaning of the espionage laws,
Title 18, USC, Secs. 793 and 794, the trans-
mission or revelation of which in any manner
to an unauthorized person is prohibited by law.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Approved For Release 20C1Mar : CIA-RDP78-03952A000100070 01,4
CONFIDENTIAL
ase II Final Report
Volume VII
EDP SYSTEM
CHIVE/R-3-65
1 March 1965
--e/t49t-Cz-e;-
DOC / REV DATE 0
ma comp OPI TYPE
DRIG CLASS PAGESgi REV CLASS
JUST NEXT REV clO AUTHi HR 104
Approved For Release 2118QajeL)8 : 0100070001-5
Approved For Release 2000/0540V
4
21AM/0100070001-5
TABLE OF CONTEI7,TS
7.1.
CHIVE EDP System Introduction
7.2.
Summary of Hardware Requirements
7
7.2.1. First Year Hardware
Pequire-
mid
ments
7.2.2. Storage Requirements
7
7.2.3. Optical Page Reader
7.3.
Files
'?
7.3.1. General
11
7.3.2. File Manipulation
1:
7.3.3. File Storage
1)
7.4. Transaction Types 3.5
7.4.1. Structure Definition
Transactions 15
7.4.2. File Maintenance Transactions 16
7.4.3. Information Retrieval
Transactions
Approved For Release 2000/05/
rvin0
;
1A-RIIP78-03952A000100070001-5
Approved For Release 2000/05/0
1;9A613a-Aai1/C1C1070001-5
7.5
Executive Control
Page
23
7.5.1.
System Requirements on the
Executive Control System
25
7.5.2.
Design Philosophy
28
7.5.3.
System Input/Output
32
7.5.4.
Task Management
35
7.5.5.
Core Storage Management
37
7.5.6.
Data Management
40
7.5.7.
Additional Executive Control
Functions
45
7.6.
Summary of Internal Data Structures
47
7.6.1.
File Format Tables
48
7.6.2.
Content of File Format Tables
49
7.6.3.
Additional File Format Table
Functions
52
7.u.4.
File Format Table Modification
53
7.6.5.
System Data File Organizations
53
7.6.6.
CHIVE System Data Pile Record
Structure
57
7.6.7.
Searching for a Value Field
62
7.6.8.
Directory File Structure
64
Approved For Release 2000/05/0
DP7rGONR
151111 t
TOIOR.0001-5
Approved For Release 2000/05/Wei.-
maw
,NTIA2.0070001-5
Page
7.7. File Maintenance 67
7.7.1. Pile Maintenance Processing 67
7.7.2. Input Transformations 63
7.7.3. Record Maintenance ("""
7.7.4. Directory File Maintenance 69
7.8. Petrieval Hit Processing 71
7.8.1. Hit Processing Functions 71
7.8.2. Structuring Commands 74
7.8.3. Editing Commands ,) 0-)
,,,
7.8.4. The Arithmetic Commands ,_1 nr
7.8.5. Writing Commands 197
7.9. Input Processing 117
7.9.1. Types of Input 117
7.9.2. Input Language Processor 117
7.9.3. Command Macros and Standing
Job Definitions 119
7.9.4. Input Job Patching 123
7.9.5. Input Functions 126
rn
Approved FQ.11%1140:1911411iM5/ r-RDP78-03952A000100070001-5
Approved For Release 2000/05/085RWRCPPIAMUME}070001-5
7.10. Output Processing
7.10.1. Output Routine Definition
7.10.2. Report-Related Processing
Functions
Page
127
127
130
7.10.3. General Report Generating
Functions 133
7.11. Security Procedures 153
7.11.1. Problem Statement 153
7.11.2. Mixed Data-Classification
Concept (All Source) 154
7.11.3. Mixed EDP Job Classification
Concept (All Demand) 156
7.11.4. Problem Solution 157
7.11.5. Decompartmentalized All-Source/
All-Service Concept 157
7.11.6. The Role of "Need-to-Know"
in CHIVE Security Design 158
7.11.7. Security Isolation of Programs
and Data in Storage 159
7.11.8. Security Isolation of Data
Transfers 161
7.11.9. Security Isolation of Jobs
and Tasks 163
7.11.10. Privileged Access to Logical
Files 165
Approved For Release 2000/05/0
,-_,WALNI(14167000,5
Approved For Release 2000/05/
?91EQkiiii&i070001-5
7.11.11.
Access to Auxiliary Storage
? Page
Units
166
7.11.12.
File Interaction and Security
Isolation
166
7.11.13.
Privileged Access to Output
Devices
169
7.11.14.
EDP Software Security
Isolation
170
7.11.15.
EDP Hardware Security
176
7.11.16.
Procedural Security
178
7.11.17.
Computer Facility Security;
Premises Security
183
7.11.18.
Information Handling Functions
and Operations
183
7.12.
Backup Procedures
187
7.12.1.
Backup Concepts
187
7.12.2.
Backup Plan
188
7.A.
User Language
Manual
192
7.A.1.
Purpose of Manual
192
7.A.2.
CHIVE Data Elements and Their
Logical Structure
193
7.A.3.
General Organization of CHIVE
Language
201
Approved For Release 2000/05
-RDP78-?33152.A0601`00670001-5
Approved For Release 2000/05/08
C.!...) 'MEM i4
DP78-03952A000100070001-5
Page
7.A.4.
Language Specifications
213
7.A.5.
Inter-Record and Inferential
Logic
262
7.A.6.
File Maintenance
279
7.B.
Input Transcription Study
285
7.8.1.
Fundamental Assumptions
286
7.B.2.
Evaluations of the Candidate
Input Systems
292
7.B.3.
Summary
321
7.C.
EDP Output Formats
335
7.C.1.
Document Index File Reports
335
7.C.2.
Vocabulary Control File Reports
343
7.C.3.
Summary Data File Reports
346
7.C.4.
Unsynthesized File Report
346
7.D.
System Macros Used in Operational
Programs
353
7.D.1.
Departure Macros
354
7.D.2.
Data Set Control Macros
357
7.D.3.
Data Set Accessing Macros
358
7.D.4.
Storage Allocation Macros
360
Approved For Release 2000/05/08s: DP7t-tInt041710fi0 001-5
CONFIDENTiloil
Approved For Release 2000/05 '.-:'ffibP78-03952A000100070001-5
7.E. Hardware Unit Characteristics
Page
361
7.E.1.
Central Processing Unit
361
7.E.2.
2361 Core Storage
371
7.E.3.
I/O Channels
372
7.E.4.
Bulk Storage Devices and
Associated Control Units
375
7.E.5.
System Communication Devices
383
Approved For Release 2000/05/08 : CIS10:078-095
P6E1911
'A'1
1-5
Approved For Release 2000/05/08 : CSECIMP78-03952A000100070001-5
FIGURES
Page
7-1 Overall View of Chive EDP System 3
7-2 Input/Output Interfaces in OS/360 33
7-3 CHIVE System Data File Record Structure 51
7-4 Matrix Representation of Terms in a
Phrase Group 59
7-5 Directory File Record Structures
7-6 Hit Processing Function
7-7 Extract
7-8 Merge P
7-9 Merge R
7-10 Append
7-11 Delete
7-12 Replace
7-1a Tally
7-14 Accumulate
7-15 Compute
7-16 Write
7-17 Copy
7-18 Explode
7-19 Print
65
72
76
78
84
88
91
93
97
100
103
108
111
114
116
Approved For Release 2000/05/08 : Clhfife78-03952A000100070001-5
Approved For Release 2000/05/0?Ectia-RDP78-03952A000100070001-5
7-20
7-21
7-22
Input Command Statements
CHIVE Input Processing
Page Arrangement - CHIVE Reports
Page
121
125
129
7-23
Samples of Line Types
135
7-24
Phrase Print Format 1
142
7-25
Phrase Print Format 2
143
7-26
Phrase Print Format 3
144
7-27
Phrase Print Format 4
145
7-28
Phrase Print Format 5
146
7-29
Phrase Print Format 6
147
7-30
Phrase Print Format 7
148
7-31
Phrase Print Format 8
149
7.A-1
Logical Structure of CHIVE Records
196
7.A-2
Sample CHIVE Record
197
7.A-3
Schematic and Tree Representation of
Document Index Record
199
7.A-4
Schematic and Tree Representation of
Information Record
200
7.A-5
Test and Command Lines on Input
Coding Sheet
204
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0VcelfA-RDP78-03952A000100070001-5
7.A-7
The Numeric OP Entry
217
7.A-8
The Relational OP Entry
218
7.A-9
Value Entries with Phrase
Coordination
220
7.A-10
Value Entries with DON'T CARE
Characters
221
7.A-11
Value Entries with Tag Occurrence
Symbol
223
7.A-13
Value Entry with $ FILE (name)
224
7.A-14
Test Line (C) Made up of Test Lines
(A,B)
226
7.A-15
Phrase Coordination in a COMPLEX
Test Line
227
7.A-16
The Uses of "c for Phrase Coordination
228
7.A-17
Test Lines with M of N Logic
229
7.A-18
Example of Use of X Operator
230
7.A-19
Summary of Test Line Rules
230
7.A-20
COPY and PRINT Command Formats
231
7.A-21
COPY and PRINT Command Lines
233
7.A-22
Test Line
235
7.A-23
COMMAND Lines for COPY and PRINT
235
7.A-24
CRITERIA Entries for Various Data
Elements
236
7.A-25
Use of CRITERIA Entry
238
7.A-26
Alternative Use of CRITERIA Entry
239
7.A-26A
DEFINE Command Line
240
Approved For Release 2000/05/%daik-RDP78-03952A000100070001-5
Approved For Release 2000/05/01EMA-RDP78-03952A000100070001-5
Page
7.A-27
Example of EXTRACT Command
242
7.A-28
Examples of WRITER Command
244
7.A-29
Examples of Append Command
245
7.A-30
Phrase Appended to a Record.
247
7.A-31
Term Appended to Phrase
247
7.A-32
Examples of DELETE Command Lines
248
7.A-33
REPLACE Command Lines
249
7.A-34
EXPLODE Command Lines
251
7.A-35
Exploded Hit Records
252
7.A-36
EXPLODE Terms Command
253
7.A-38
TALLY Command Lines
254
7.A-39
TALLY and APPEND Result for Output
256
7.A-40
Command Lines for Accumulating Across
a File
259
7.A-41
Command. Lines for Accumulating TALLY
Result
258
7.A-42
COMPUTE Command Lines
259
7.A-43
COMPUTE n Referenced in an APPEND
Command Line
261
7.A-44
Enriched Querying
264
7.A-45
MERGER with Single and with Multiple
Merge Keys
265
7.A-46
Input Records Before MERGER Command
266
7. 47
MERGED RECORD with Common CNC 20
268
7.1\.-48
Input Records with Common Values for
Multiple Merge Key PNO and CNC
269
Approved For Release 2000/05/08?MIRDP78-03952A000100070001-5
Approved For Release 2000/05/08stg*TRDP78-03952A000100070001-5
7.A-49
Merged Record with Common Multiple
Page,
Merge Key
270
7.A-50
Test and Command Lines for Inferential
MERGER Logic
272
7.A-51
MERGEP with Single and Multiple Merge
Keys
274
7.A-52
Phrases with Common Values for Merge
Keys
275
7.A-53
Matrix of VALUES for Phrases 1-5
276
7.A-54
New Phrase Merged from Phrases 1,2,
and 3
277
7.A-55
New Record after MERGEP Execution
278
7.A-56
MERGE Phrases for Multiple Merge Keys
280
7.A-57
Command Lines for File Maintenance
APPEND
281
7.A-58
Command Lines for File Maintenance
DELETE
282
7.A-59
Command Lines for File Maintenance
REPLACE
283
7.B-1
Operating Costs Punched Card Systems
294
7.B-2
Operating Costs Paper Tape Systems
298
7.B-3
Operating Costs Paper Tape Systems
299
7.8-4
Operating Costs CRT/KB Systems
308
7.B-5
Operating Costs CRT/KB Systems
309
7.B-6
Operating Costs Farrington Page
Reading System
316
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/OFF.8R-RDP78-03952A000100070001-5
Page
7.B-7
Operating Costs Ravinow (CDC) Page
Reading System
317
7.B-8
Operating Costs Recognition Equipment
Page Reading System
318
7.B-9
Comparative Operating Costs of Candi-
date Input Systems (off-line)
323
7.C-1
Document ID-Term Report-Cover Page
336
7.C-1
Document ID-Term Report-First Page
337
7.C-1
Document ID-Term Report-Body Page
332
7.C-2
Abbreviated Header-Contents Report
340
7.C-3
Full Header-Contents Report
341
7.C-4
KWIC Title Report
342
7.C-5
Map Catalog Card
344
7.C-6
Organization/Facility Control Report
34
7.C-7
Leader Appearance-Source Citation
Report-Cover Page
347
7.C-7
Leader Appearance-Source Citation
Report-Body Page
340
7.C-8
Condensed Leader Appearance Report
249
7.C-9
Individual Travel to Foreign Countries
Report-Cover Page
350
7.C-9
Individual Travel to Foreign Countries
Report-Body Page
351
Approved For Release 2000/05/08 SECIAERDP78-03952A000100070001-5
Approved For Release 2000/05/08 : Cltapf'78-03952A000100070001-5
TABLES
Page
7-1 File Interaction Security Transition
Table 168
7.A-1 Summary of Rules for Command Lines 211
7.B-1 Cost Summary of Candidate Input Systems 324
7.B-2 Summary of Requirements for Typists/
Keypunch Operators and Proofing Clerks
for Candidate Input Systems 326
7.B-3 Comparison of Alternative Input Trans-
cription Methods 328
7.B-4 Subsequent Tasks for OCR and OCS After
Choosing the Page Reader Approach 330
7.B-5 TentativP Page Reader Specifications -
Project CHIVE 331
Approved For Release 2000/05/08 : CLMET78-03952A000100070001-5
Approved For Release 2000/05/08 WEF1DP78-03952A000100070001-5
Chapter 7.1.
CHIVE EDP SYSTEM INTRODUCTION
The EDP portion of CHIVE will perform the following
functions:
- Build and maintain files
- Search files
- Create sub-files from existing files
- Display data
The CHIVE system must process many different data files.
The files range from the large document index files to
relatively simple and small decode files. Further, it
is necessary to provide the capability to add new files
to the system and also to revise the structure of existing
files. All of the above considerations have dictated a
general approach towards the CHIVE EDP design; that is,
a single, flexible set of programs will perform the
functions outlined above on any of a large class of files.
The programs will be implemented in a fashion that will
permit new files to be introduced to the system and old
files to be restructured with no change to the programs
themselves.
SYSTEM INTRODUCTION
7.1.
-1-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/0SECDRE1CIA-RDP78-03952A000100070001-5
The feature that gives flexibility to the system is
extensive use of parameters. All of the CHIVE programs will
rely on external descriptions of the data and transactions
to be processed. The data descriptions taken collectively
are called the File Format Tables. Each table describes a
file and its constituent items. Included in the description
are record structure, types of item, storage, addressing
rules, and legality parameters. By changing the item
descriptions, the user of the system can make the file(s)
reflect current needs. It is possible to create a file
and then through experience gained from usage, revise the
file structure for increased usefulness. Input transactions
to the CHIVE system are also broken down into individual
functional elements or parameters. The net result is that
the maintenance and retrieval operations can be easily
controlled by non-EDP users.
An overview of the CHIVE EDP system is shown in
Figure 7-1. The system is divided into five major modules:
- CHIVE Executive
- File Maintenance
- Retrieval
- Report Generation
- Format Table Maintenance and File Restructuring
SYSTEM INTRODUCTION
7.1.
-2--
Approved For Release 2000/05/08': CIA-RDP78-03952A000100070001-5
SECRET
Approved ForRelease2000/05/08 : CIA-RDP78-03952A000100070001-5
Figure 7-1
Output
OVERALL VIEW OF CHIVE EDP SYSTEM
Nnput
(OS/360 )
CHIVE
Executive
Report
Generation
A
Retrieval
Maintenance
Data
Files
File
Structuring
6, Restruc-
/tur'ni
-3-
FFT s
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Approved For Release4CM5/08 : CIA-RDP78-03952A000100070001-5
The CHIVE executive module performs these functions:
it interfaces with Operating System/360, it analyzes all
CHIVE inputs and switches control to the proper operating
module, and bookkeeps the status of all machine files under
CHIVE control.
The File Maintenance module creates and updates the
CHIVE files. It consists of generalized routines for
processing files and special purpose routines for handling
raw input. Depending on the type of file to be processed,
t will operate in either a demand or batch mode.
The Retrieval module accepts all CHIVE search requests
and translates them into search routines. Control is then
passed to these generated programs which then search the
indicated file. The records selected can be subjected
to additional manipulation which can result in the creation
of a sub-file or a completely restructured file. This
special file manipulation, called Hit Processing is fully
described in Chapter 7.8.
The Report module accepts any file in the CHIVE system
and prints it according to specified parameters. The
system allows the user considerable latitude in formatting
reports. Direct control is permitted as to the sequence
SYSTEM INTRODUCTION
7.1.
-4-
Approved For ReleasgaM/05/08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/Ncitk-RDP78-03952A000100070001-5
of presentation, the amount of printing, which items
should be printed, and their placement and spacing.
The Format Table module generates file format tables
and updates them. It also restructures existing files,
when their associated format tables have been revised, to
insure compatibility with the new tables.
Inspection of Figure 7-1 should make it clear that
any processing of a CHIVE data file is always performed in
conjunction with its associated file description. This
feature is the primary device used in attaining system
flexibility. One other important aspect of the system,
which is difficult to show pictorially, is the fact that
the'CHIVE system operates on a demand basis in a multi-
programming environment. It will actually be under
control of the IBM-provided operating system. Thus,
the CHIVE programs will operate concurrently with other
computer jobs completely unrelated to CHIVE.,
The 'functional capabilities of the EDP system are
controlled by the user via a command language. A
description of this language is presented separately in
Appendix 7.A. as a user's manual. The information in
7.A. provides the non-EDP reader with the essence of
the design goals of the CHIVE EDP system. Also this
SYSTEM INTRODUCTION
7.1.
-5-
Approved For Release 2000/05/OgicM-RDP78-03952A000100070001-5
Approved For Release 200Wegt# : CIA-RDP78-03952A000100070001-5
section allows all readers to determine the extent of
generality built into the system.
SYSTEM INTRODUCTION
7.1.
-6
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08 IZ1AERDP78-03952A000100070001-5
Chapter 7.2.
SUMMARY OF HARDWARE REQUIREMENTS
The CHIVE Initial System will be implemented on
the OCS IBM System/360, Model 60. CHIVE will be
operated in a multiprogrammed mode with other OCS
jobs. The IBM Operating System/360 will provide the
executive control and multi-programming facilities for
the CHIVE program modules.
7.2.1. FIRST YEAR HARDWARE REQUIREMENTS
At the end of one year operational test period, CHIVE
will
require the utilization of
the
following equipments:
Qty.
Units
(%)
Use Estimated
1
2060 - Central Processor
15%
1
1403 - Printer
13%
1
1402 - Reader-Punch
1%
8
2400 - Magnetic Tapes
5%
1
2311 - Disk
13%
1
Console
1
Optical Page Reader
*The optical page reader and console utilization is not
shown since specific selections have not yet been made.
HARDWARE REQUIREMENTS
First Year
7.2.1.
-7-
Approved For Release 2000/05/08 :sBODP78-03952A000100070001-5
Approved For Release 20CBTA06878 : CIA-RDP78-03952A000100070001-5
The percent utilizations given for the System/360
are based on 100% utilization being equal to 176 hours
per month. Further, these estimates were based on the
estimated inputs, processing required, and outputs of the
reference and information files projected for the CHIVE
System. The type of activities that were used to make
these estimates were searching, sorting, maintenance,
record building, input processing and output processing.
These utilization figures do not contain any time
allocated to file backup generation or processing. File
backup was left out due to the lack of definition in
this area.
The times for the central processor include estimates
for all the transactions mentioned previously plus
overhead time for the Operating System/360.
7.2.2. STORAGE REQUIREMENTS
A gross volume breakdown of the CHIVE storage needs
is shown in the table below. Units are expressed in
millions of characters.
1st Year
2nd Year
3rd Year
Permanent Disk
45
29
70
Temporary Disk
10
10
10
Data Cell (2321)
0
112
212
HARDWARE REQUIREMENTS
Storage Requirements
7.2.2.
-8-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08 if--kDP78-03952A000100070001-5
It is anticipated that all of the storage requirements
for the first year will be assigned to disks. A 2321 Data
Cell Drive (400 million characters rated capacity) will be
required for CHIVE storage at the beginning of the second
year. During the second and third years, CHIVE storage
will be divided between the 2321 Data Cell Drive, disks
and the 2400 Magnetic tapes.
It should also be pointed out that all of the above
estimates do not contain time for conversion, maintenance
or operation of any inherited files such as SR or IRS.
7.2.3. OPTICAL PAGE READER
In reaching a decision about the type of input
transcription system to be used to input data and queries
to the CHIVE EDP system, four alternatives were examined
in great detail: a punched card system, a punched paper
tape system, a cathode ray tube-keyboard system, and a
page reader system. The study considered the
usefulness, flexibility, adaptability, speed,
etc., of these four approaches. On the basis
cost,
dependability,
of cost-
effectiveness considerations and simplicity of operation,
the page reader system was selected as most responsive
to immediate and long range CHIVE requirements. A
detailed discussion of the findings of the study and the
methodology employed is in Appendix 7.B.
HARDWARE REQUIREMENTS
Optical Page Reader
7.2.3.
-9-
Approved For Release 2000/05/08 ?spcliEltDP78-03952A000100070001-5
Approved For Release 2000/05/08qE4KIRDP78-03952A000100070001-5
Chapter 7.3.
.FILES
7.3.1. GENERAL
The CHIVE EDP system is designed to service a large
number of files. There are two characteristics that are
common to all files. First, they are all externally
described in file format tables. This feature allows
a single set of programs to process any files in the
system. Second, the structure of the records is such
that complex relationships found in documents can be
retained. Simpler structures such as those found in
card files are also encompassed by the system.
The basic characteristics of the record structure
are: sets or groups of values can be linked together.
These sets can contain varying combinations of descriptors
and in addition the constituent descriptors can occur
within a set many times if desired. A set of descriptors
and their values when linked together are called a phrase.
Several individual phrases can be stored in a record.
The descriptors in a file can be subdivided if desired,
into mutually exclusive phrase types. It is permitted to
FILES
General
7.3.1.
-11-
Approved For Release 2000/05/08:aleOP78-03952A000100070001-5
Approved For Release 200VE48r: CIA-RDP78-03952A000100070001-5
store several different types in a record. Uhen it is
desired to control a file with unique record identification,
a record control phrase is defined. This phrase is different
than other phrase types since it can occur only once in CI
record. Also, the particular descriptor(s) in the phrase
that are chosen to provide the unique designation can occur
but once.
7.3.2 FILE MANIPULATION
All CHIVE files are described in external tables.
These tables are used to drive maintenance, retrieval anC
report programs during the processing of a filo. Files
can be initiated in either of two methods. 'no first is
Lc manually provide a file description to the system.
Prom this point, it is now possible to build a file from
raw input data. The second method is to create files frroT,
those already in existence. In this case, the original file
format table is revised. The revision is sometimes per-
formed automatically; and, on other occurrences, tie
revisions must stem from manual inputs.
7.3.3. FILE STORAGE
Provision is made for serial and random storage. Files
stored serially will be allocated to magnetic tapes. Random
FILES
Pile Stora7-
7.3.7.
-12-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Cir
Approved For Release 2000/05/0W:c8R-RDP78-03952A000100070001-5
files will be stored on disks and data cells. Those files
stored in random fashion will have directory files to
facilitate accessing. It should be noted that the IBM
provided software allows the files to be I/O device independent
Thus, no special programming effort is required if the
storage media of a file is changed.
Storage requirements for CHIVE files have been grossly
estimated for three one-year periods. Period one is
associated with the first year of operational test concerned
25X6 . Period two includes the remainder of the Far
East while Period three includes the Far East and USSR
operations.
The most important files or types of files in the
initial system are briefly described below. Their
projected volumes and storage allocations are also
indicated.
Document A.ndex File (DIF) - contains descriptions of
documents residing in CHIVE hard copy files.
Term File (TF) - contains the values of terms in the DIF
and the associated storage addresses.
Decode Files (DF) - contain machine stored configurations
and desired output equivalents.
FILES
File Storage
7.3.3.
-13-
Approved For Release 2000/05/08iRODP78-03952A000100070001-5
Approved For Release 2001:Magt : CIA-RDP78-03952A000100070001-5
Information Files (IF) - contain data on named objects or
events associated with named objects, the data itself is
organized, evaluated, and non-redundant.
Access Files (AF) - contain information necessary to access
randomly stored files. Logically the AF are similar to the
TF.
CHIVE Files-Volumes (in millions of characters) and
Associated Storage Devices
NAME
I
II
III
DIF
36.0
90.0
170.0
Disk
Data Cell
Data Cell
TF
21.0
54.0
Disk
Disk
DF
.65
1.0
1.0
Disk
Disk
Disk
IF
17.0
18.0
18.5
Tape
Tape
Tape
AF
.2
.5
1.5
Disk
Disk
Disk
FILES
File Storage
7.3.3.
-14-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/054gr-RDP78-03952A000100070001-5
Chapter 7.4.
TRANSACTION TYPES
This Chapter summarizes the following types of
transactions in the CHIVE EDP system:
- Structure definition transactions
- File maintenance transactions
- Retrieval transactions
7.4.1. STRUCTURE DEFINITION TRANSACTIONS
There are three major types of data structures
which must be defined to the CHIVE EDP system:
- Input formats
- File Format Tables
- Output report formats
In the initial CHIVE system all of these data
structures will be defined at the machine language
level by assembler or compiler language definitions. The
File Format Tables generated by the assembler/compiler
process will be treated as data sets which would be
catalogued via the Operating System/360 in appropriate
files. The structure definitions of inputs and report
outputs may contain instructions as well as parameter
TRANSACTION TYPES
Structure Definition
7.4.1.
-15-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000$1E4ter: CIA-RDP78-03952A000100070001-5
tables and hence would be catalogued as programs in the
system program library.
7.4.2. FILE MAINTENANCE TRANSACTIONS
File maintenance transactions can be categorized by
three factors for internal EDP processing. The first
factor will be the type of file organization and storage:
(a) sequentially accessed tape files, (b) sequentially
accessed files on direct access storage devices, and
(c) directory accessed files. The second factor will be
the type of record specification: (a) specific and
(b) conditional. The specific maintenance transactions
identify a single record by providing the unique record
control terms. Conditional record specifications involve
test conditions based on record content which define one
or more records for a single maintenance transaction to
act upon.
The third factor differentiates between externally
generated transactions and the special transactions
generated internally to maintain the directories for
directory access files. These directory files are
maintained by transactions which are automatically
generated by the system during the processing of maintenance
transactions for a directory accessed file.
TRANSACTION TYPES
File Maintenance
7.4.2.
-16--
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08-TRDP78-03952A000100070001-5
These three factors define seven categories of
maintenance transactions on the basis of internal EDP
processing requirements. These seven transaction
types, their characteristics, and the requirements for
handling and internal processing are listed below:
7.4.2.1. File Maintenance Transactions with Conditional
Record Specification for Sequential Tape Master Files
- Normally run in limited size batches
- Sequential record search
- File copied with changes
of the file
onto a new tape
- Each file record compared with the record
criteria of all transactions in the batch.
7.4.2.2. File Maintenance Transactions with Specific
Record Control Terms Specified for a Sequential Tape
Master File
- Normally run in batches
- Transactions must be ordered by record control
within the batch
- Not necessary to limit batch size
- Sequential record search of the file
- File copied with changes onto a new tape
- Transaction batch passed in parallel with master
file.
- Only one record control from the transaction
batch compared with each file record
TRANSACTION TYPES
File Maintenance
7.4.2.2.
-17-
Approved For Release 2000/05/08 .?RI!i,EfDP78-03952A000100070001-5
Approved For Release 200MIL018 : CIA-RDP78-03952A000100070001-5
7.4.2.3. File Maintenance Transactions with Conditional
Record Speancation for Sequentially Accessed Files on
1-537gEE Access Storage Device
- Normally run in limited size batches
- Sequential record search of the file
- Only the records changed need to be rewritten
onto the storage device
- If space is not available, the
relocated and chained into the sequence
record will be
- Each file record compared with the record
specification of all transactions in the batch
/.4.2.4. File Maintenance Transactions with Specific
Record Control Terms Specific for a Segue-Fin-ally Accessed
File on Direct Access Storage Device
- Can be run one at a time or in batches
- File search using OS/360 indexed sequential access
method
- Only a small number of file records need to be
accessed and compared with record criteria
- Only the records changed need to he rewritten onto
the storage device
- If space is not available, the record will be
relocated and chained into the sequence
7.4.2.5. File Maintenance Transactions with Conditional
Record Specification for Directory Accessed File
- Can be run one at a time or in batches
- Directory access file search using the directory
file
TRANSACTION TYPES
File Maintenance
7.4.2.5.
-18-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0EfftrAIRDP78-03952A000100070001-5
- Only a small number of master file and directory
file records need to be accessed and compared
with record criteria
- Only the master records changed need to be rewritten
on the storage device
- The required changes to the directory file are
generated as separate maintenance transactions on
the directory file
7.4.2.6. File Maintenance Transactions with Specific
Record Controls for Directory Accessed Files
- Can be run one at a time or in batches
- Directory access file search using the record
control directory file
- Only a small number of master file and record
control directory file records are accessed and
compared with record search criteria
- Only the master records changed need to be rewritten
on the storage device
- The required changes to the record control
directory file are generated as transactions
for these directory files
7.4.2.7. File Maintenance Transactions for Directory Files
- Generated during the processing of master record
maintenance on directory accessed files
- Executed in the same manner as specific maintenance
transactions to sequentially accessed files on
direct access storage devices (Section 7.4.2.4. above)
7.4.3. INFORMATION RETRIEVAL TRANSACTIONS
Information retrieval transactions involve three
data handling operations: record retrieval, hit record
processing and output generation. There are three classes
TRANSACTION TYPES
Information Retrieval
7.4.3.
-19-
Approved For Release 2000/05/08 :sq68DP78-03952A000100070001-5
Approved For Release 2000WW: CIA-RDP78-03952A000100070001-5
of retrieval commands which indicate the operational
character of the transaction.
7.4.3.1. Retrieval Type Commands
The COPY and EXPLODE commands both retrieve records
or portions of records and copy them directly onto
output files. These files may be used subsequently as
further input files within the transaction or as final
output of the transaction.
The EXTRACT command retrieves a single record or
portion of a record into a staging area for subsequent
processing.
7.4.3.2. Processing Type Commands
APPEND, DELETE, REPLACE, and MERGER commands
process individual records which are EXTRACTed into a
work area. The MERGER command will combine records in a
file on the basis of common terms.
TALLY, ACCUM, and COMPUTE are internal processing
functions which result in output eventually, either in
a specified report or automatically if not otherwise
specified.
TRANSACTION TYPES
Information Retrieval
7.4.3.2.
-20-
Approved For Release 20090/A8 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/05EMIRDP78-03952A000100070001-5
7.4.3.3. Output Type Commands
The WRITER and WRITEP commands form a record from
data in an EXTRACT staging area and place it in an output
file.
The PRINT command causes a designated file to be
printed according to the parameters specified in the
command.
TRANSACTION TYPES
Information Retrieval
7.4.3.3.
-21-
Approved For Release 2000/05/08: CIA-RIQP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08s.E6FAIRDP78-03952A000100070001-5
Chapter 7.5
EXECUTIVE CONTROL
The CHIVE programming system will be composed of a
number of interacting programs that are continually
vying for the resources of the computer. The CHIVE Executive
Control System (ECS) brings general overall program manage-
ment and control to this environment. This section of the
410.
report attempts to summarize all pertinent facts relating
to the present status of executive control philosophy and
design in CHIVE.
Traditionally, monitors or executive control sections
di of serial programming systems have been looked upon as
being second class citizens--their main functions were
to pass parameters from one program to the next and to
woo perform some common I/O activites. The advent of
sophisticated hardware interrupt structures and multi-
programming have antiquated this viewpoint. Executive
mai control development in CHIVE, rooted in the later
philosophy,has undertaken three shifts in the point of
mor
view as to how such a multiprogrammed system should be
implemented. The present implementation approach and
UMW
EXECUTIVE CONTROL
7 . 5 ?
-23-
Approved For Release 2000/05/08 :siattliDP78-03952A000100070001-5
Approved For Release 2000$064: CIA-RDP78-03952A000100070001-5
the one used in the discussions in this report is that
of a modified IBM System/360 operating system. All
terms in this section of the report are consistent with
Operating System/360 terminology.
To put the discussions on the nunerous facets of
executive control in perspective, formal and informal
requirements and underlying philosophies are covered
first. General features and historical design considera-
tions are also presented in order to provide background
for the body of the report--Executive Control Functions.
Some of the Executive Control reports used in the
preparation of this paper are:
CHIVE/N-2-64 Types of Multiprogramming
CHIVE/N-3-64 Preliminary Executive Control
Flow Chart Exnlanation
CHIVE/N-5-64 Preliminary Macro Definitions
CHIVE/N- P-6?4 Executive Control - Transaction
Management
CHIVE/N-9-64 Core Storage Management
CHIVE/N-16-64 General Functional Characteristics
of the Executive Control Subsystml
E,LCUTIVE
7.5.
-24-
Approved For Release 200kft?igl : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/0PWRDP78-03952A000100070001-5
CHIVE/N-1-65 I/O Management
CHIVE/W-28-64 Preliminary Functional Design of
the Executive Control Subsystem
7.5.1. SYSTEM REQUIREMENTS ON THE EXECUTIVE CONTROL SYSTEM
This section contains a compiled set of formal and
informal requirements levied on the Executive Control System
by other parts of the total CHIVE system. This list of
requirements were extracted primarily from CHIVE/W-17-64,
however, comments from other reports and meetings have also
been used when appropriate.
7.5.1.1. Turn-Around Time Constraints
The most stringent requirement for turn-around time
is given for priority information queries. In summary,
this requirement reads: One priority information query
can be expected every thirty mintues in the initial
system, and it should be processed in two minutes
(W-17-64). This seemingly innocuous requirement
interjects a strong flavor of real-time into the system.
This differs from other requirements which point to a more
conventional system.
7.5.1.2. Priorities
The system must be able to handle jobs having various
processing time constraints on them. These time limita-
EXECUTIVE CONTROL
System Requirements
7.5.1.2.
-25-
Approved For Release 2000/05/08 sOfefDP78-03952A000100070001-5
Approved For Release 2001WM : CIA-RDP78-03952A000100070001-5
tions are inherently contained in the "priority" of the
job. At present, three levels of priority have emerged
in the system--priority, normal, and background.
Undoubtedly more levels will emerge as design continues.
A priority specification will be assigned externally
(explicitly) to every job that enters the system. Internal
priorities based on these external priorities will also
be used within the executive control system to schedule
jobs for execution.
7.5.1.3. Input Processing
The primary requirement for the ECS in this area is
to accept input jobs into the system on demand. This
concept of "demand processing" calls for jobs to be
read in and stacked on the input staging unit. They will
be picked up in priority sequence from the staging unit
when they can be executed.
7.5.1.4. System Logs
In Chapter 5.5. of this report a number of logs are
discussed. The ECS will be called on to handle and
monitor many of these logs.
7.5.1.5. Operator Communications
In a multiprogrammed system, the operator must be
made aware of the status of the processing. This does
-26-
EXECUTIVE CONTROL
System Requirements
7.5.1.5.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
4010
Approved For Release 2000/05/08 WElkOP78-03952A000100070001-5
not imply that continuous on-line printouts are required,
but the operator should be kept informed on significant
internal happenings. For example, advance notice of
job requirements would be given, such as the mounting of
required storage devices.
7.5.1.6. Non-CHIVE Jobs
It is expected that CHIVE and non-CHIVE jobs will
make concurrent use of the same computer. Within the
structure of a multiprogrammed system this dual type
of operation is certainly possible. Since this is a
recent requirement, its full import on CHIVE requirements
has not been ascertained.
7.5.1.7. Checkpoint/Restart
Some minimal capability is required for this
programming function. Two types of procedures will be
used. '.1:?A ECS will execute macros from operational
programs to provide a means of checkpointing and
restarting. Checkpoints will include the core storage
contents associated with the transaction requesting the
checkpoint, as well as designated files. Long operational
programs on the other hand can be written in such a way
that checkpoints of their data files can be performed
by the program itself.
EXECUTIVE CONTROL
System Requirements
7.5.1.7.
-27-
Approved For Release 2000/05/08 :R1-8.DP78-03952A000100070001-5
Approved For Release 2006?93% : CIA-RDP78-03952A000100070001-5
7.5.1.8. Backup Processing
There should be a minimum capability for using CHIVE
programs and files on a backup computer. This implies
the existence of a minimal executive control system. All
basic functions should be provided through the ECS for a
backup system, but this capability should not require
extensive additional design and reprogramming activities.
7.5.1.9. Interpreting/Parsing
Requests to the CHIVE system will be written in a
specially designed user language. This language will
have to be interpreted, parsed and compiled into computer
instructions. Some means of integrating this activity
into the structure of the executive control system is
required.
7.5.1.10. Security
Security considerations are discussed in Chapter 7.11.
Security in the EDP system will have to be handled for
the most part through the ECS.
7.5.2. DESIGN PHILOSOPHY
An overall philosophy of CHIVE Executive Control has
evolved since work began in May of last year. This
philosophy is based on CHIVE requirements, on a general
EXECUTIVE CONTROL
Design Philosophy
7.5.2.
-28-
Approved For Release 200EM8 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/08 MiikERDP78-03952A000100070001-5
understanding of current computer hardware and software
capabilities, and on professional experience and
intuition. In a sense, this philosophy serves as a
bridge between the CHIVE sstem requirements and reasonable
executive control functions. The main points of ,this
philosophy are listed and discussed below.
7.5.2.1. Multiprogramming
Multiprogramming in CHIVE is a necessity to
effectively handle the computer's resources as well as
to adequately meet CHIVE processing requirements. An
executive control system designed along multiprogramming
lines attempts to share all of the resources of the
computer (CPU control, storage, I/O communication paths)
between jobs in varying stages of execution. This
implies that some jobs will have to be queued while they
are awaiting the use of a temporarily busy or an over-
loaded resource. System job control is, then, a major
function of the Executive System.
7.5.2.2. I/O Bound
Because of the large volume of external storage
required in the CHIVE system, communication paths to
external storage units will necessarily be heavily used.
EXECUTIVE CONTROL
Design Philosophy
7.5.2.2.
-29-
Approved For Release 2000/05/08 :sia-gDP78-03952A000100070001-5
Approved For Release 200eggian : CIA-RDP78-03952A000100070001-5
This condition implies that the CPU will at times be
idle while data is being transferred to or from the
external units. Since this condition is recognized as
a strong possibility, steps should be taken in all
phases of design work to use CPU and I/O resources in
an optimum manner.
7.5.2.3. Demand Processing
Even though some of the present turn-around time
requirements are very tight, the CHIVE system does not
have the overall characteristics of a real-time system.
However, since input data will be read in immediately
upon receipt, it can be termed a "Demand Processing"
system.
7.5.2.4. Modular Design
In the basic design of a comprehensive operating
system, simplicity and flexibility must be paramount
objectives. These characteristics can best be realized
by using the concept of modularity in the design of the
programming components of executive control. Programming
considerations are reduced and growth potential is
increased with such a design goal.
EXECUTIVE CONTROL
Design Philosophy
7.5.2.4.
-30-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08MIRDP78-03952A000100070001-5
7.5.2.5. Macro Oriented
Since CHIVE's multiprogramming system separates
programming responsibilities into executive control and
operational, effective communication between all parts of
the system must be maintained. This is accomplished
through an extensive set of system macros. By using
these macros all requests from operational programs are
channeled to the appropriate executive control programs
which perform the required activity. See Appendix 7.D.
for a further discussion of system macros.
7.5.2.6. Programming Services
A number of additional programming services can be
provided in the system because of the centralization of
control that will exist in the Executive Control System.
Such features as debugging aids, checkpoint capability
and diagnostic interfaces are possible and desirable
through the comprehensive structure of the Executive
Control System
7.5.2.7. Bulk Core Storage
Bulk core storage is useful in the CHIVE computer.
Bulk core is similar to regular core except that it has a
slower access speed and comes in blocks of considerable
EXECUTIVE CONTROL
Design Philosophy
7.5.2.7.
-31-
Approved For Release 2000/05/08aIRDP78-03952A000100070001-5
Approved For Release nocgmet : CIA-RDP78-03952A000100070001-5
size. The main reasons for using bulk core in CHIVE
are to:
- Reduce or eliminate reading of programs from
external storage units.
- Reduce or eliminate roll-out and roll-in.
(See Section 7.5.6.2.)
- Enable temporary results to remain in core rather
than in data sets on direct access devices.
(See Section 7.5.7.)
7.5.3. SYSTEM INPUT/OUTPUT
The input/output functional area in OS/360 is
described diagrammatically in Figure 7-2. The modified
OS/360 version will be harmonious with this structure.
Possible changes may be necessitated by the CHIVE priority
and input language decomposition requirements.
EXECUTIVE CONTROL
System Input/Output
7.5.3.
-32-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
ENE.
IPPR
IP,
Approved For Release 2000/0APEalA-RDP78-03952A000100070001-5
System
Input
Devices
Console
Figure 7-2
INPUT/OUTPUT INTERFACES IN OS! 360
Job Scheduler
1--11111.
Reader/
Interpreter
Master Scheduler
Inp<
Staging
Output
Staging
<
Job Scheduler
Initiator/
Terminator
Other Parts of OS/360
and
Operational Programs
-33--
Output
Writer
Approved For Release 2000/05/08 : CIA-RDF'78-03952A000100070001-5
SECRET
System
Output
Devices
Approved For Release 2000/ISEMETCIA-RDP78-03952A000100070001-5
7.5.3.1. Input Processing
The major functions performed at this level are:
- Reading input control information onto the input
staging device.
- Error checking and tabulation of job control
information.
- Allocation of resources to jobs.
- Scheduling of job steps to process each jOb through
the system.
- Handling all correspondence from the console
(operator).
OS/360 will provide appropriate and adequate input
programs to perform all phases of input processing. The
Job Scheduler is composed of two parts which handle input
processing. The Reader/Interpreter will read "on demand"
input from all system input devices and queue them into
priority sequence on the random access input staging
disk. Then it will create a Job Control Table and allocate
resources before the job is scheduled for execution. The
second section of the Job Scheduler will pick up the jobs
from the input staging device and sets up a task control
block (TCB) for each task. The Master Scheduler specifically
handles all correspondence to and from the console.
7.5.3.2. Output Processing
When individual tasks are being executec . within a
job, generated output will be written on the random access
EXECUTIVE CONTROL
System Input/Output
-34- 7.5.3.2.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0eaT-RD1278-03952A000100070001-5
output staging device. The output will be accumulated in
this way until an "end of output" signal is given to the
Terminator (output end of Job Scheduler). The Output
Writer is relayed the necessary information, and the data
on the output staging device will be serially written on
the appropriate system output devices.
The revisions to OS/360 in the area of Output Processing
will be primarily in the incorporation of remote output
terminals to the system. This modification in the initial
version of OS/360 will vary in complexity depending on
whether common carrier facilities are used.
7.5.4. TASK MANAGEMENT
After a job is entered into the system and the
necessary preliminary activities are performed by the
Job Scheduler, subsequent processing of that job is
handled by the Task Dispatcher and the Task Supervisor.
Task Management will dynamically supervise and monitor
all tasks that are in the system. This includes such
functions as:
- Selection of tasks for execution.
- Maintaining the status of the program related queues
- Servicing of all types of interrupts (except I/O)
- Maintaining logs of important system events.
- Providing communication services between programs.
EXECUTIVE CONTROL
Task Management
-35-
Approved For Release 2000/05/08 :,fiV,RepP78-03952A000100070001-5
Approved For Release 2000MHT: CIA-RDP78-03952A000100070001-5
OS/360 provides adequate support in the area of
task management.
7.5.4.1. Task Dispatcher
The dispatching procedure centers around the
selection and preparation of tasks for execution. The
more important steps in this procedure are:
- Selection of the task at the top of the Ready
Queue for execution.
- Requesting of core storage and reading in of
programs from the system library (if the programs
are not already in core).
- Performing dynamic program relocation and assigning
storage protection keys (if not already done).
A requirement may emerge to alter the rule by which the
Ready Queue is ordered. Also some modifications in the assign-
ing of storage protection keys may be required to meet
security specifications.
7.5.4.2. Task Supervisor
The main function of the Task Supervisor is to monitor
the execution of tasks. This supervision is characterized
by the following:
- Handling all types of communication between
programs.
- Routing all interrupts to their respective
service programs.
EXECUTIVE CONTROL
Task Management
7.5.4.2.
-36-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0?MAIRDP78-03952A000100070001-5
- Maintaining system queues, lists, and blocks.
- Performing standard and ad hoc logging of system
events and activities.
Except for the possible exception of some specialized
types of logs, all the above functions will be performed
by OS/360.
7.5.5. CORE STORAGE MANAGEMENT
The purpose of core storage management is to:
- Allocate and release core storage blocks on
request.
- Maintain control over all temporary core storage
areas.
- Resolve conflicting needs for core storage.
The handling of I/O buffers by OS/360 might be
changed to give a better utilization of core by
assigning I/O buffers just prior to I/O command
execution time rather than at data set open time. Core
storage usage could be reduced by requiring all programs
to be reentrant (a reentrant program is so written that
more than one task can use it concurrently) so that system
will operate more efficiently when core is overburdened.
In OS/360 tasks and some supervisor functions are
the entities that are assigned core storage. The main storage
supervisor in its general control over core storage maintains
EXECUTIVE CONTROL
Core Storage Management
7.5.5.
-37-
Approved For Release 2000/05/08 ikl,t1DP78-03952A000100070001-5
Approved For Release 2000ftrefift: CIA-RDP78-03952A000100070001-5
interrelated references to a task's storage areas. These
references take the form of special lists and keep track
of storage blocks assigned to a task as well as free
areas within these storage blocks. (0S/360 assigns
storage in multiples of 2K byte blocks)
7.5.5.1. Storage Macros
GET MAIN is the macro used to request core storage
from OS/360. Requests can be of three types. The amount
of space allocated by type of request in OS/360 is as
follows:
Types of Requests One Block Multiple Block
Request Request
Non-Conditional---- Block of size Each block of a
Must have specified specified specified size
amount.
Conditional--Special Block of size
return if request specified
cannot be met.
Each block of a
specified size
Variable--Special
return if minimum
length is not
available.
Largest block
available within
specified limits
If the request is for problem program storage (task
storage), the request is filled if possible with available
storage from the task's previously assigned blocks.
Otherwise, groups of 2K blocks are allocated to the task
EXECUTIVE CONTROL
Core Storage Management
7.5.5.1.
-38-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
wad
Approved For Release 2000/05WEA-RDP78-03952A000100070001-5
to help meet the request. Other types of storage are
handled similarly, except for read only program areas.
These are handled differently because of future uses that
can be made of their storage areas.
FREE MAIN is the macro instruction to dynamically
release main storage, so that it becomes available for
future use. The return procedure is checked for
recognizable errors. If the return frees one or more 2K
byte blocks, the blocks are returned from the task storage
to overall free storage. A check is made to see if the
2K block could fulfill a pending roll-out or is needed
for a roll-in.
7.5.5.2. Roll-out and Roll-in
When storage is requested and no space is available,
Core Storage Management tries to free space occupied by a
program (read only) that is not being used. If this fails,
Core Storage Management will acquire the space occupied
Jill by one or more tasks by writing (rolling out) the contents
on a direct access device. The decision to roll-out a task
is made on the basis of task priority.
Roll-in (reading these sections back into core)
takes place automatically as soon as the original space is
available again. If conflicts over core use occur often,
EXECUTIVE CONTROL
Core Storage Management
7.5.5.2.
-39-
Approved For Release 2000/05/0?E&WRDP78-03952A000100070001-5
Approved For Release 2000813/013T: CIA-RDP78-03952A000100070001-5
it may be desirable in OS/360 to have data sections
organized in a way they could be rolled into areas other
than the original area.
7.5.6. DATA MANAGEMENT
A multiprogramming system such as that envisaged for
CHIVE implies that the I/O Supervisor will take on more
functions than the traditional IOCS. Brief definitions
of those terms used in this section are as follows:
- Data Sets: a named collection of data (previously,
this was called a logical file)
- Volume: a physical piece of hardware used for
storing data sets (previously, this was called a
physical file)
- Catalog: a hierarchical structure of indexes of
data sets names for locating data sets.
The additional functions (beyond those of conventional
IOCS systems) included in OS/360 are:
- Allocation of space on direct access devices.
- Expanded library and catalog organization including
automatic location of data sets by name alone.
- More device independence given in part by the
capability to defer specifications such as buffer
size, blocking factor, device identification, and
in some cases device type until the job is submitted
for actual execution.
- Wide variety of supported data organizations.
EXECUTIVE CONTROL
Data Management
7.5.6.
-40
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05WFUIA-RDP78-03952A000100070001-5
- Protection of data sets by:
- Not executing I/O orders for a data set
outside of its extent, i.e., its logical and
physical boundaries.
- Protection of a classified file by
external "password" or key.
use of an
- Protection against concurrent updating
same record by different tasks.
of
The protection of classified files by use of a
password requires operator action in OS/360. If it is
desired to have an internal security tag on a data set
rather than on record or lower level, an internal data
set password, handled completely internal to the
programming system, would be more desirable and would be
incorporated in modified OS/360.
7.5.6.1. Data Set Location Control
The functions of controlling space on direct access
devices and of finding and keeping track of data sets are
covered in this section.
Every direct access volume has a volume table of
contents used to keep track of its data sets and free
space. When a user desires to create a data set on the
volume, Data Management assigns the data set space from
the volume's free space, and enters a data set label for
this data set in the volume table of contents. If the
EXECUTIVE CONTROL
Data Management
7.5.6.1.
-41-
Approved For Release 2000/05/064-RDP78-03952A000100070001-5
Approved For Release 2000/'#T CIA-RDP78-03952A000100070001-5
data set is to be cataloged, additional work as described
in the following section is required.
Cataloging of a data set permits its storage and
retrieval by name alone, otherwise the volume number
would be needed. Generated data sets are easy for the
user to reference, because the system automatically
recatalogs them when a new member if added. Also a
set of control statements and macro instructions are
provided for maintenance of the library and catalog.
7.5.6.2. Data Set Organization and Use
OS/360 will probably support the access methods
and data set organizations presented below. There are
three record formats:
- Fixed-length logical records
- Variable-length logical records
- Record of unspecified characteristics (each
physical record is treated as one block without
regard to internal structure)
7.5.6.3. Data Set Organization
Four data set organizations will be available. In
the sequential organization each record in a data set is
related to others by its position.
The indexed sequential method applies to any direct
access device and provides sequential and random accessing.
EXECUTIVE CONTROL
Data Management
7.5.6.3.
-42-
Approved For Release 200MA8 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/0Vqa-RDP78-03952A000100070001-5
A data set is created in sequence of both physical location
and collating sequence. If subsequent insertions cannot
be put into proper sequential location, they are placed
in a separate area and chained in logical sequence.
Cylinder and track indexes provide random accessing of
information.
The direct organization for direct access devices
lets the user specify where each record is located. A
relative track address is used as a starting point of
search for the referenced record.
The partitioned data set organization lets the
programmer collect groups of sequential records on a
direct access device for rapid retrieval through a
directory within the data set.
7.5.6.4. Data Set Accessing Languages
The two supported langauge levels are Basic and
Queued. The Basic level is for physical record accessing;
the Queued is for logical records. Queued describes the
automatic queuing of input and output records on buffers.
The supported access methods are identified as follows:
Data Organization Types Queued Basic
Sequential QSAM BSAM
Partitioned XXX BPAM
Indexed Sequential QISAM BISAM
Direct XXX BDAM
EXECUTIVE CONTROL
Data Management
7.5.6.4.
-43-
Approved For Release 2000/05/08EqkbRDP78-03952A000100070001-5
Approved For Release 2000/etrAfT. CIA-RDP78-03952A000100070001-5
OS/360 may provide for user provided machine language
access methods and record formats.
7.5.6.5. I/O Buffers
Buffers are located in user core and are allocated
either at assembly time or dynamically at execution time.
Techniques for their use are:
- Simple Buffering: Buffer belongs only to data
set using it.
- Cyclic Buffering: Buffers for physical records
are shared; and ownership may pass from one data
set to another, depending on use.
- Exchange Buffering: Buffers for logical records
are shared between input and output data sets.
Buffers belonging to data sets of a task as in
OS/360 simplify the I/O Supervisor. This does'not give
as good core utilization as dynamic assignment of
buffers just prior to I/O command execution.
7.5.6.6. OPEN and CLOSE Macros
The OPEN macro operates on a Data Control Block (DCB)
that refers to the desired data set. The DCB provides
data set descriptive information such as name, organization,
language, buffering techniques, and location. This
information is provided by the program symbolic deck,
Data Definition Statement (from the Job Control Language),
EXECUTIVE CONTROL
Data Management
7.5.6.6.
-44-
Approved For Release 200W8 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/0?E.C8R-RDP78-03952A000100070001-5
and the Data Set Control Block. The CLOSE macro restores
the DCB to its original condition prior to open.
7.5.7. ADDITIONAL EXECUTIVE CONTROL FUNCTIONS
In addition to the primary functions discussed above,
several secondary functions can be easily implemented due
to the amount of control inherent in 05/360. A few of the
more important extras incorporated into OS/360 are discussed.
7.5.7.1. Debug Supervisor
Since there are many activities being performed
concurrently in a multiprogrammed system, the necessity
for a central control over program errors is paramount.
Increased debug services can be offered through the
operating system to the operational programmers. These
services provide a set of debug macros unparalleled in
conventional operating systems.
7.5.7.2. Checkpoint/Restart Supervisor
Most programs that require a long execution time make
periodic checkpoints of part or all of their intermediate
results. These checkpoint records would be initialized or
restarted if an error occurred in later stages of processing.
In addition to this standard practice, OS/360 will provide
EXECUTIVE CONTROL
Additional Functions
7.5.7.2.
-45-
Approved For Release 2000/05/08Eq1A-RDP78-03952A000100070001-5
Approved For Release 2800,RE108 : CIA-RDP78-03952A000100070001-5
the user, upon request, dynamic checkpoints of the areas
in core that are related to his programs and data.
7.5.7.3. Diagnostic Supervisor
Even on today's highly reliable equipment, malfunctions
still occur. Many of the hardware errors can be analyzed
and isolated by special programming packages called
diagnostic programs. OC/360 has several types of diagnostic
programs built into its structure. Whenever a hardware
failure has been recognized, OS/360 will call in automatically
the appropriate diagnostic program. If possible after the
reason for failure has been found, normal processing will
be continued with the malfunctioning hardware blocked off
until it can be repaired.
EXECUTIVE CONTROL
Additional Functions
7.5.7.3.
-46-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0PWRDP78-03952A000100070001-5
CHAPTER 7.6
SUMMARY OF INTERNAL DATA STRUCTURES
The CHIVE system data files will contain data
organized according to the following well defined
structure:
- Term: A term is the basic logical data unit.
This is made up of two parts, a tag and a value.
The tag defines the "term type" or use in the
system and the value is the actual data of the
term.
- Phrase: A phrase is made up of two or more
terms linked together to form a meaningful unit
of data. (In some cases, phrases with only
one term can occur.)
- Phrase type: Two types of phrases are defined
in the CHIVE system: header phrases and content
phrases.
- Header Phrase: A CHIVE record contains exactly
one header phrase.
- Content Phrase: Any number of content phrases
can occur in each record of a CHIVE file.
- Phrase Group: A set of content phrases in a
CHIVE record terms of the same types,
i.e., with the same tags.
m.
- Record Control Terms: Each record a CHIVE
file will have one or more terms which taken
together contain unique data value(s) in the
file. The record control terms therefore uniquely
identify each record. There can be only one occurrence
of these record control values in a record.
SUMMARY OF DATA STRUCTURES
7.6.
-47-
Approved For Release 2000/05/08 Ed4EFDP78-03952A000100070001-5
Approved For Release 20?M138 : CIA-RDP78-03952A000100070001-5
7.6.1. FILE FORMAT TABLES
The format of each CHIVE system data file is defined
by a File Format Table (FFT). The File Format Table
describes the format and content of each system data file
and provides an interface between the CHIVE files and
problem programs. The declaration of file content and
structure via an FFT frees CHIVE programs from consideration
of a particular file. Program development and implementation
need be oriented only towards the general classification
of structures which can be represented in an FFT and the
processing techniques which can be applied to these
structures.
The FFT also provides a mechanism by which a program
can define an intermediate file to subsequent programs
requiring access to the intermediate file.
7.6.1.1. FFT Jurisdiction
The employment of File Format Tables in CHIVE will
be restricted to the definition of system data files.
System processing files (e.g., program and subroutine
libraries), directory files, and OS/360 input and output
staging files are excluded from FFT definition.
The FFT will provide the problem programs, which
are not oriented towards a particular CHIVE file,
sufficient information to process any system data file.
SUMMARY OF DATA STRUCTURES
File Format Tables
7.6.1.1.
-48-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/04-RDP78-03952A000100070001-5
7.6.2,, CONTENT OF FILE FORMAT TABLES
File Format Tables will contain various structural
definitions and process controls in a tabular format. The
structural definitions outlined below will be carried
primarily at the tag level, supplying an entry point to
a record for each tag declared eligible in the file.
7.6.2.1. Phrase Level of Tag
Indicates the phrase group to which the tag belongs.
7.6.2.2. Occurrences of Tag Allowed in a Phrase
The number of terms which can be represented by the
tag. The eligible occurrence modifiers are one (1) and
variable (V).
7.6.2.3. Occurrences of Tag Reguired in a Phrase
The requirement parameter specifies if the appearance
of a term with the tag is manatory or optional.
7.6.2.4. Term Value Length
The number of characters which can be employed to
represent the value of the term in the record. This
parameter consists of the maximum and minimum number of
characters allowed.
SUMMARY OF DATA STRUCTURES
Content of Format Tables
7.6.2.4.
-49-
Approved For Release 2000/05/088RpieDP78-03952A000100070001-5
Approved For Release 2001rEtaitC18 : CIA-RDP78-03952A000100070001-5
7.6.2.5. Value Location Within EDP Record
The addresses required to calculate the location
of the values for the tag in each record. The values of
fixed-length terms which appear will be referenced by a
single displacement vector pointing to the high order
position of the value, relative to the starting position
of each record.
The values of all other terms not meeting the above
criteria will be located by the displacement vectors
listed below. Figure 7-3 and Section 7.6.6. describe
their use in accessing all the terms within a specified
record.
- An address vector in the FFT specifying the
relative location of a Term Locator Control
Block. This TLCB is located in the same posi-
tion in each record of the file.
- The Term Locator Control Block of each record
contains a second displacement vector locating
a Value Locator Block in the record servicing
the tag of interest. The position of the
VLB will vary from record to record.
- A second displacement vector in the FFT locates
the position in the Value Locator Block which
contains the actual displacement vector to the
values of all terms with the particular tag in
the record. The second FFT vector is required
because the Value Locator Block will specify the
actual displacement vectors for multiple tags.
SUMMARY OF DATA STRUCTURES
Content of Format Tables
7.6.2.5.
-50-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0PW-RDP78-03952A000100070001-5
Figure 7-3
CHIVE System Data File Record Structure
Dedicated Fields
RECORD LENGTH
RECORD CONTROL
TAG 3
TAG 4
TAG i
Count
TAG 1
TAG 2
' Value
Value
Value
Value
Value
PHRASE GROUP MASK LENGTHS
P.G. 2
Length
Length
P.G.
'Length
TERM LOCATOR CONTROL BLOCKS
TLCB 1
TLCB 2
TLCB k
TLCB hp.
Vector
to
VLB 1
Vector
to
VLB 2
?
Vector
to
VLB k
TAG 5 - HEADER PHRASE TAG
# of Values
Length 1
Length 2
Length n
Value I
Value 2
Value n
TAG 6 - PHRASE GROUP 1 TAG
# of Values
Length .1
Length 2
Length n
Mask 1
Mask 2
Mask h
Value I
Value 2
Value n
TAG 7 - PHRASE GROUP 2 TAG
# of Val.
Length 1
Length 2
Length n
Mask I
Mask 2
Mask n
Value 1
Value 2
Value n
TAG M - PHRASE GROUP j
# of Values
Length 1
Length n
Mask 1 Value 1
Mask n Value n
IVALUE LOCATOR FILLDS
VLB 1
VLB 2
1 *VLB k
Vector
to
TAG 5
Vector
to
TAG 6
Vector
to
TAG 7
Vector
to
TAG in
-51-
Approved For Release 2000/05/08 : EAFP7 8-03952A00010007 0001-5
Approved For Release 201311ME8 : CIA-RDP78-03952A000100070001-5
7.6.2.6. Mask Length Location
The address of the field in the record which contains
the phrase mask length for the phrase group of which the
tag is a member.
7.6.3. ADDITIONAL FILE FORMAT TABLE FUNCTIONS
Special FFT services required by problem programs
cannot be completely identified until program design has
progressed to a more advanced state. A preliminary
outline of required process controls include the following:
- Identification of tags employed as record and
phrase controls.
- Tags dedicated to contain record security classi-
fication and dissemination.
- Identification of the directory files servicing
the system data files.
- An indication if the values for the tag are
eligible for conditional maintenance.
- Justification procedures for each tag, indicating
the method of aligning variable length values for
comparison. The modifiers include:
- Right justification
- Left justification
- Floating justification: a special character
in the value is to be employed as a reference
point.
- An indicator for each tag specifying if the tag
is represented in a directory file.
SUMMARY OF DATA STRUCTURES
Additional Functions
7.6.3.
-52-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08 'WiTRDP78-03952A000100070001-5
- Subroutine or table declarations for each tag whose
values require special processing for conversion
from external to internal format. The declaration
will be employed by retrieval, unconditional and
condition maintenance programs.
- General value characteristics, which at a minimum
specify if a value is in binary coded decimal,
packed decimal, or binary representation.
7.6.4. FILE FORMAT TABLE MODIFICATION
The CHIVE retrieval programs will have some capa-
bility to modify the data structure and content of an
existing system data file. The modified files produced
by retrieval are candidates for inclusion as system
data files. In addition to the modified file, retrieval
produces a File Format Table specifying the structure
of the revised file, which must be included in the
system library if the revised file is to become a
permanent member of the CHIVE EDP system.
File revision requirements not within the purview
of retrieval must be processed as new file creations,
with a program assigned to accomplish the required
transformation. The supporting File Format Table must
be generated and introduced into the CHIVE library as
a new file format definition.
7.6.5. SYSTEM DATA FILE ORGANIZATIONS
The organization of the system data files for
storage and retrieval is dependent on the storage devices
SUMMARY OF DATA STRUCTURES
File Organizations
7.6.5.
-53-
Approved For Release 2000/05/08 : gimp P78-03952A000100070001-5
Approved For Release 200SMC(8 : CIA-RDP78-03952A000100070001-5
and the expected activity of the files. The files will
be organized either as ordered sequential files on tapes
or direct access storage devices, or as directory
accessed files stored on direct access storage devices.
CHIVE/Nr1-65 describes the variots standard file organi-
zations and accessing methods provided by OS/360. The
sequential files will be physically ordered by their
record control terms. Sequential files stored on direct
access storage devices will be processed using the OS/360
indexed sequential access method. This technique provides
direct access to records through index tables which
relate the record control terms to the physical location
of the records. The processing of these files will use
routines and macros which are a standard part of the
Operating System/360.
7.6.5.1. Sequential Files on Tape,.
Files stored on magnetic tape will be ordered on
the terms designated for record control. File main-
tenance transactions, and brand new records will be
processed in batches defined as jobs to the Operating
System/360. Unconditional maintenance transactions
in these batches will be sorted on the record control
terms into the same order as the file records prior
SUMMARY OF DATA STRUCTURES
File Organizations
7.6.5.1.
-54-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08S.EMIRDP78-03952A000100070001-5
to the actual maintenance processing. The original
file and the maintenance batch will then be processed
sequentially against each other using the OS/360 sequential
access method. A new updated file will be produced on
another tape. Logical maintenance transactions can be
processed in the same batch as the specific unconditional
maintenance transactions by appending a blank sort key
to each conditional transaction before ordering the batch.
They can be grouped together and matched against all the
file records.
7.6.5.2. Sequential Files on Direct Access Storage Devices
Files may also be stored on direct access storage
devices for sequential access. The records in these
files will be ordered on the record control terms and
stored sequentially using the OS/360 indexed sequential
technique. This will permit direct access to specific
records for maintenance and output. Specific file main-
tenance transactions can be processed individually even
if inputted as batches, or in batches ordered by record
control. Logical maintenance transactions or retrieval
transactions require a complete search of the file and
should be processed in batches for efficiency to avoid
multiple passes against the file.
SUMMARY OF DATA STRUCTURES
File Organizations
7.6.5.2.
-55-
Approved For Release 2000/05/08 :sEMDP78-03952A000100070001-5
Approved For Release 20M0510-8 : CIA-RDP78-03952A000100070001-5
7.6.5.3. Directory Accessed Files
Directory accessed system data files will consist
of three distinct segments or subfiles. The main seg-
ment will contain the master data records in the order
they are received by the system. The other two segments
will contain directory files. One directory file will
provide access to the records through the record control
terms and the other through access terms designated in
the File Format Table. These access terms can be selected
from the header and content phrases of the master data
records.
The directory files will be stored on direct access
storage devices using the OS/360 indexed sequential
access method. The records in the directory file will
consist of a term and a set of postings, one posting
for each phrase which contained the term. Each posting
will contain a storage address for the master record and
a phrase number to show which phrase in the record
contained the term. The records in the directory file
will be ordered by the tag and the specific value.
Changes and additions to the directory files will be
automatically generated from maintenance and record
addition transactions processed against the system data
SUMMARY OF DATA STRUCTURES
File Organizations
7.6.5.3.
-56-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0eal:RDP78-03952A000100070001-5
file. No changes can be specifically directed against
the directory files.
7.6.6. CHIVE SYSTEM DATA FILE RECORD STRUCTURE
CHIVE system data file records will have the same
general record structure for all types of physical
storage devices. The data terms and required control
data will be contained in fixed length and variable
length fields. The tags can be partitioned into arbitrary
tag groups by the user, according to their frequency
of co-occurrence. These tag groups do not have to be
related in any way to the phrase group(s) to which the
tags belong. The phrase linkages of terms are
represented by phrase mask fields which are associated
with each individual term value. This mask in each
record will have a single binary bit position dedicated
to each content phrase of the phrase group to which the
term belongs. A mask bit position set to one indicates
that the term value was contained in the phrase represented
by that position. A bit position set to zero indicates
that the term was not contained in that phrase. The
phrase masks from all the terms in a phrase group
would form a coincidence matrix of binary bit positions
with each row representing a term and each column a
SUMMARY OF DATA STRUCTURES
Record Structure
7.6.6.
-57-
Approved For Release 2000/05/08 :Sichf4ERDP78-03952A000100070001-5
Approved For Release 20680/1108 : CIA-RDP78-03952A000100070001-5
phrase. This coincidence matrix concept is illustrated
in Figure 7-4.
The records in a system data file will have two
major sections as shown in Figure 7-3--a dedicated
section and an optional section. The dedicated
section of a record will be made up of fixed length
data and control fields which will have the same posi-
tion and space allocated to each such field in all
records of the file.
dedicated data
fields
be stored in the File
The position and length of the
and internal control fields will
Format Table. The optional section
of the records will be made up of fixed and variable
length data fields and control fields. If they do not
occur in a particular record of the file no storage space
is allocated to them. The position and length of the
optional data fields is stored in Value Locator Blocks
in the optional section of the record. The position of
these Value Locator Block fields is, however, stored in
Term Locator Control Block fields located in the dedicated
portion of the record.
7.6.6.1.
This
the total
Record Length
control field contains a binary count of
record length in bytes. This controls the
-58-
SUMMARY OF DATA STRUCTURES
Record Structure
7.6.6.1.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08 WgDP78-03952A000100070001-5
Figure 7-4
MATRIX REPRESENTATION OF TERMS IN A PHRASE GROUP
Logical Phrase Group
Phrase Number
Terms
1
Al
P1
Cl
D1
2
A2
B2
Cl
D2
3
A3
B2
C2
D3
4
A2
C2.
DI
5
A3
B3
C2
DI
6
Al
B2
Cl
D2
Bi Ci, Di, = Term with t g A? B, Cy or D and value i
e rms
Al
A3
Bl
B2
B3
Cl
C2
DI
1)2
D3
Phrase Numbers
12
'
5
6
78
1
0
o
0
0
0
o
o
1
0
1
0
0
0
o
o
0
o
o
o
o
1
1
0
0
0
0
1
mi0
0
0
0
0
0
0
0
0
0
a::
0
1
i
0
0
0
i
0
0
i
0
0
0
0
0
so
0
0
0
0
5 9 -
Approved For Release 2000/05/08 : Cykliipp78-03952A000100070001-5
Approved For Release 20Vafig8 : CIA-RDP78-03952A000100070001-5
storage of the record on the storage devices and
defines its length for processing routines.
7.6.6.2.Record Control Fields
These fields contain the value of the record control
terms. They are located in their ranking order immediately
following the record length fields. They are used to
order sequential files and to uniquely identify a single
record for file maintenance operations.
7.6.6.3. Dedicated Data Fields
These fields contain the values of fixed length
data terms which are required to appear in every
record. Their location is given by a displacement vector
specified directly in the File Format Table.
7.6.6.4. Phrase Group Mask Length Fields
These control fields contain a binary count of the
actual number of phrases in each phrase group of a
particular record. The actual length of the phrase
mask field will be the minimum number of bytes
required to store this number of bits.
7.6.6.5. Term Locator Control Blocks
The term types or tags will be partitioned into tag
groups according to the expected co-occurence of the
SUMMARY OF DATA STRUCTURES
Record Structure
7.6.6.5.
-60-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
MI
_
Approved For Release 2000/05/MMIA-RDP78-03952A000100070001-5
data terms in records. A block will be designated in
the record for each tag group which has been defined in
the File Format Table. Each block will contain a 16-bit
binary vector address showing the relative location of the
Value Locator Control Block in the record if any terms
with a tag belonging to the particular tag group are
present in the record. It will contain all zero bits if
mid
no tag of its tag group is present in the record. A
ami special TLCB will also be reserved for tags not defined in
the FFT which are appended to the record during retrieval
"Hit Processing." This occurs when retrieval processing
is not under FFT Control,
me 7.6.6.6. Value Locator Blocks
These blocks will normally occur as the last fields
of a record. Their presence in the record is indicated
dio by the presence of a non-zero vector address in the
associated Term Locator Control Block. The length of
each Value Locator Block will be indicated in the File
amo Format Table. A Value Locator Block will be made up of
a set of 16-bit Value Locator Fields, one for each tag
in the greup of tags controlled by the block.
Each Value Locator ?Field in the Value Locator Block
SUMMARY OF DATA STRUCTURES
Record Structure
7.6.6.6.
-61-
Approved For Release 2000/05/0%eaftRDP78-03952A000100070001-5
Approved For Release 200@?.18 . CIA-RDP78-03952A000100070001-5
will contain either all zeros if no terms with the
particular tag are present in the record, or a location
vector showing the relative location of the Value Count
Control Field for the set of values for terms with that
tag in the record.
7.6.6.7. Optional Value Fields
The value fields will be partitioned by term type
tag. The individual values for a particular tag will be
ordered in /360 collating sequence within the partition
for that tag. Each partition will be headed by a Value
Count Control Field showing the number of values with this
tag in the record. Each value field will be
preceded by a Value Length Control Field showing the
number of bytes of data in the value field. If the value
field is for a term in a phrase group that can occur more
than once, it will also be preceded by a Phrase Mask Field.
Terms appearing in a phrase group that can occur only
once per record will have an entry made up of a Value
Length Field, and a Value Field.
7.6.7. SEARCHING FOR A VALUE FIELD
The dedicated value fields will be searched directly
by obtaining their relative location vector and length
SUMMARY OF DATA STRUCTURES
Searching for a Field
7.6.7.
-62-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08 PEARDP78-03952A000100070001-5
from the File Format Table. The optional value fields
will be searched indirectly. This access will use the
File Format Table entries and control fields in the
record as described below.
Step 1: Using the File Format Table extract the
relative location in the record of the Term Locator
Control Block for the particular tag. Also extract the
location in the Value Locator block of the Value Locator
Field for the particular tag (for use in Step 3).
Step 2: Extract the relative location of the Value
Locator Block within the record from the Term Locator
Control Block.
Step 3: Extract the relative location of the Value
Count Field from the Value Locator Field in the Value
Locator Block.
Step 4: Search through the individual value fields
using the Value Count Field, the Value Length Fields
and the Phrase Mask Length to control the search stepping.
A search for a particular value would involve match
comparisons against all the value fields. A search
for a term value in a particular phrase would also
match against the particular bit position representing
that phrase in the phrase mask.
SUMMARY OF DATA STRUCTURES
Searching for a Field
7.6.7.
-63-
Approved For Release 2000/05/0%tagTRDP78-03952A000100070001-5
Approved For Release 2CKE0285108 : CIA-RDP78-03952A000100070001-5
7.6.8. DIRECTORY FILE STRUCTURE
The directory files will have a very simple
structure. The Record Control Directory File and the
Term Director File are the same in overall file organiza-
tion. They will be stored and accessed using the Operating
System/360 indexed sequential technique. They are defined
and processed as distinct file types, however, because,
they have different record structures (see Section 7.7.5.).
7.6.8.1.
The
Term Directory File Record Structure
Term Directory File records will be variable
length. The record structure is shown in Figure 7-5.
The first field in the record will be a record length
field containing a binary number defining the record
length in characters or bytes. This will be followed
by the term entry consisting of a one byte code for
the tag, a value length field and a variable length value
field. The record
record address and
will be 32 bits or
postings will consist of a 24 bit
an 8 bit phrase number. Each posting
one /360 computer word. For
processing ease the first posting will be located in the
first complete computer word following the term entry.
From one to three blank bytes will be inserted in the
record following the value field to make the length of
the record count and the term entry length an even
multiple of four bytes.
SUMMARY OF DATA STRUCTURES
Directory File
-64- 7.6.8.1.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0gW-RDP78-03952A000100070001-5
Figure 7-5
DIRECTORY FILE RECORD STRUCTURES
Term Directory File Record
Record
Tag
Val.
Value
Length
Lgth
-
Posting 1
Posting 2
c:
,
c6rd
cation
Ph.
No.
Record
Location
Ph.
No.
Posting (n)
Record
tocatia
Record Control Directory File Record
Record Cortrol Fields
Record
Location
-65-
Ph.
No.
Approved For Release 2000/05/08 :sqek-gDP78-03952A000100070001-5
Approved For Release 2008S1:5/Er8 : CIA-RDP78-03952A000100070001-5
7.6.8.2. Record Control Directory File Record
The record control directory file records will be
fixed in length. The record control terms which are
located in the dedicated section of the system data
file record are fixed length. There will be only one
record address for each record of this file. This
structure is also shown in Figure 7-5.
SUMMARY OF DATA STRUCTURES
Directory File
7.6.8.2.
-66-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08 F4TIDP78-03952A000100070001-5
Chapter 7.7.
FILE MAINTENANCE
7.7.1. FILE MAINTENANCE PROCESSING
File maintenance processing covers two major
activities, adding new data and changing data. The
three types of file organization each present a separate
problem in maintenance. Tape files must be copied to
change records. Therefore file maintenance transactions
should be batched for economy. The indexed sequential
files on direct access storage have a direct access
capability using the record control terms. These files
can be changed without copying all the records onto
another storage unit. New records can be chained
into the sequence by reserving extra space throughout
the file for this purpose. The file can then be copied
onto a tape and reloaded when this chaining becomes too
cumbersome. Directory accessed files must also have
concurrent directory file maintenance transactions
generated whenever a system data file change affects a
directory file entry. These transactions cannot be
completed until the system data file record is stored
in its storage device.
FILE MAINTENANCE
Processing
7.7.1.
Approved For For Release 2000/05/08 : gMipP78-03952A000100070001-5
Approved For Release 2090(1)6/08 : CIA-RDP78-03952A000100070001-5
Record changes, i.e., adding, deleting and replacing
terms or phrases in records, can be classed in two ways
according to the method of specifying the record to be
changed. Specific maintenance transactions use the
record control terms to spe4fy an exact record.
Adding complete records is always classed as specific
maintenance. Conditional maintenance transactions use
some combination of test conditions to specify one or
more records to be changed. Conditional maintenance
transaction which indicate the specific record to be
changed but conditionally specify the phrases or terms
to be updated will be handled as specific changes.
7.7.2. INPUT TRANSFORMATIONS
Transcript sheets or input records for adding
records to files will be processed at input time into
CHIVE format records. Special preprocessor routines will
be developed to process the various types of inputs for
each CHIVE file. The processing of these records will be
controlled by the job specification card (see Appendix 7.A.3.)
which will be transferred to the maintenance processing
routines with the records.
7.7.3. RECORD MAINTENANCE
Record maintenance, which involves the modification
of a single record after this record is found by the
FILE MAINTENANCE
Record Maintenance
-68- 7.7.3.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08 ?t?AElkDP78-03952A000100070001-5
appropriate search mechanism will be similar for all files.
The location of the data to be added, deleted or Changed
will be determined with record search routines. These
routines use the File Format Table, to find a particular
data field. After the field is deleted inserted or
changed the necessary changes will be made in the Record
Length Control Field, Tag Locator Blocks and Value
Locator Blocks to complete the operation. After all the
required changes have been made to the record it will be
stored back in the file.
7.7.4. DIRECTORY FILE MAINTENANCE
Directory file maintenance transactions are generated
whenever a change is made to a term in a directory accessed
file which has been flagged in the File Format Table as an
access term. A term posting deletion is generated if a
term is deleted, a term posting addition if a term is
added and both an addition and a deletion if a term is
changed. These transactions will be processed as soon
as the record has been returned to the file. . If a
complete record is added to the file, a posting addition
transaction for each access term in the record is
generated before the record is stored. These transactions
will be processed as soon as the system data record has
FILE MAINTENANCE
Directory File
7.7.4.
-69-
Approved For Release 2000/05/08 : Lt-A9P78-03952A000100070001-5
Approved For Release 20GEIERNO8 : CIA-RDP78-03952A000100070001-5
been stored in its file. If a record is deleted, one
or more posting deletion transactions are generated and
processed. The record is flagged and written back into
the file and its location saved to be used for storing
new records.
If a posting addition is processed for a term
which has not previously been entered in the directory
file, the term will be added to the directory file and
an entry made on an output history tape to be used by
the file analyst to check new terms for possible errors.
If a delete posting transaction is processed which
deletes the last record posting of an access term, the
term will be deleted. An entry is made on the history
tape and the disposition notice showing that the term
is no longer used in the file.
FILE MAINTENANCE
Directory File
7.7.4.
-70-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08 SeGIORDP78-03952A000100070001-5
Chapter 7.8.
RETRIEVAL HIT PROCESSING
The hit processing in the CHIVE EDP system involves
those record modification actions taken after specified
search criteria have been met, but before any output
processing occurs. The relation of hit processing to
the other major elements of a retrieval transaction are
illustrated in Figure 7-6.
7.8.1. Hit Processing Functions
Hit processing functions are classified into four
categories:
- Structuring Commands
- Extract
- Merge Records
- Merge Phrases
- Editing Commands
- Append
- Delete
- Replace
- Define
HIT PROCESSING
Functions
7.8.1.
-71--
Approved For Release 2000/05/08 : C4HAP78-03952A000100070001-5
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Figure 7-6
HIT PROCESSING FUNCTION
Inter-
mediate
Output
Test
Conditions
Search
Output
Prepara-
tion
Hit
Processing
Machine
Output
-72--
Report
Output
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05MMIA-RDP78-03952A000100070001-5
- Arithmetic Commands
- Tally
- Compute
- Accumulate
- Writing Commands
- Copy
- Explode
- Write
- Print
In general a command will include one or more
operands:
Data - such as terms, sort keys, print fields,
computed results, etc.
Files - either input or output files.
Search Criteria - such as the term, phrase or record
specification for a "hit."
Limit values - a numeric value to indicate the maximum
number of times that a command is to
mei be performed.
mai
VINO
Appendix 7.A. gives a complete description of the
command syntax and symbol conventions.
HIT PROCESSING
Functions
7.8.1.
-73--
Approved For Release 2000/05/ClayRDP78-03952A000100070001-5
Approved For Release 20014 : CIA-RDP78-03952A000100070001-5
7.8.2. STRUCTURING COMMANDS
7.8.2.1. EXTRACT
COMMAND LIMIT SEARCH CRITERIA FILE DATA
Record Phrase Term I 0
(Ident)
EXTRACT (Number) (Ident) (Ident) (Name)
(Ident) (Ident) (Ident)
Command Description: EXTRACT causes the operand(s)
specified in the Search Criteria to be written into a
system generated work area for subsequent manipulation.
No explicit reference to the work area is necessary. Its
presence is assumed by the system.
The data elements that may be specified are terms,
phrases, or whole records. No restrictions are placed
upon the type of element that may be extracted.
The EXTRACT command operates upon records only and
is the entry point for the inner loop of the Hit Processor.
It exists principally to preserve records or record
fragments for additional processing. Specific reference
must be made to the input file upon which the command is
to operate.
HIT PROCESSING
Structuring Commands
7.8.2.1.
-74-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0PW-RDP78-03952A000100070001-5
The Limit feature will cause the EXTRACT operation
to cease when the number of elements extracted equals
the Limit number.
- Processing Logic: See Figure 7-7 EXTRACT Logic
7.8.2.2. MERGE Phrases
COMMAND LIMIT SEARCH CRITERIA FILE DATA
Record Phrase Term I 0
MERGEP - (Key)
Command Description: MERGEP causes all phrases, in
the same record, containing a common value (or values)
for the tag (or tags) specified as the merge key, to be
combined into a single phrase. The merge key is entered
in the Data field. This command operates within a given
record; never across records, and then only upon the
record elements present in the EXTRACT work area.
- Processing Logic: See Figure 7-8 MERGEP Logic
7.8.2.3. MERGE Records
COMMAND LIMIT SEARCH CRITERIA FILE DATA
Record Phrase Term I 0
MERGER (Name) (Name)(Key +
Result
C & L
Tag)
HIT PROCESSING
Structuring Commands
7.8.2.3.
-75-
Approved For Release 2000/05/08s:EekhRDP78-03952A000100070001-5
Approved For Reliige2001:1165/08 : CIA-RDP78-03952A000100070001-5
_Extract
Store
Return
File
Name
N Equal
Branch to
Error
Routine
-76 -
Approved For Release 20 0/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Figure 7-7
Extract
Move Hit
Element
into
Work Area
Branch
to
Edit
Garman:I
Routine
-77-
Increment
Extract
Counter
4 Branch
to
)
Return
Approved For Release 2000/05/OgkiR-RDP78-03952A000100070001-5
Approved For Release 2000/05/08UCREIRDP78-03952A000100070001-5
Figure 7-8
Merge P
Branch
to
Return
Store
Return
Load
Merge
Keys
Copy Header
Phrase Grcup
into Result
Area
Set up Ccntent
Matrix in
Result Area
And All Row
Masks for
Exact Match
- .--
Approved For Release 2000/05Th/08 : CIA-RDP78-03952A000100070001-5
SECRET
Branch
to
Return
Approved For ReleaseARRN?
?
?
Merge P
b1A-RDP78-03952A000100070001-5
Flag
Duplicates
Multi Key to
ulti Key )
Routine
Approved For Release 2000/05/
Reproduce
Phrase Mask
into Result
Record
--7 9 --
CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/088BEWERDP78-03952A000100070001-5
cfure 7-
Merge P
or Phrase
Masks
Together
Reproduce
Combined
Mask into
Re suit Reoard
Branch
to
More
Value
Finish Up
Routine
Branch
to
Return
Branch
to
Write
Routine
-80-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/0?/WE61A-RDP78-03952A000100070001-5
Figure 7-8
Multi
Key
Routine
Merge P
And Row
Masks for A
Multi Tag
Key
Merge Key Value
Combination Occurs
in One or More
Phrases
Flag
Combine
Phrases
or Phrase
Column
Masks
Together
Reproduce
Combined
Cdurm Mask
into Result
Record
Branch
to
Finish Up
-81-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08 SifIlRDP78-03952A000100070001-5
Figure 7-8
Merge P
Scan Carbine
Flag Row for
Blanks
Branch
to
Finish
Up
- 82-
Branch
to
and Row
Reproduce
Original
Colima Phrase
Mask into
Result Record
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
OW
ar0
Approved For Release 2000/05/01301M-RDID78-03952A000100070001-5
The MERGER command causes all records, in a named
input file, which contain a common term (or terms) to be
physically co-located in a named output file. One or more
tags may be specified as the merge key in the Data field.
The MERGER command operates across records, but
within a given file. Its action is similar to MERGEP
but it operates on a broader boundary.
As in the MERGEP, the result of the MERGER is a
record (or several records) which combine common terms
within a searchable entity that were previously not
coordinated.
- Processing Logic: See Figure 7-9 MERGER Logic
7.8.3. EDITING COMMANDS
7.8.3.1. APPEND
COMMAND LIMIT SEARCH CRITERIA FILE DATA
Input Output
APPEND (Number) (Ident) - - (Data)
(Ident) (Ident)
Command Description: APPEND causes the data elements
specified in the DATA field to be added to the phrase or
record specified by the search criteria. The resultant
HIT PROCESSING
Editing Commands
7.8.3.1.
-83-
Approved For Release 2000/05/08s:EeleDID78-03952A000100070001-5
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Figure 7-9
Merge R
,,/"MergeR
< Command
N
Store
Return
Load All
Command
Operands
I Set Up
I Output
File
Input
'File Nam
Equal e
?
(s5
Branch
to
Return
Branch to
Error
Routine
\RecoCheck /
rd
?
(1T
Scan for
Tag Match
Approved For Release 200005/W: CIA-RDP78-03952A000100070001-5
Approved For Release 2000//015545741AgRDP78-03952A000100070001-5
Merge R
Store
Return
0
Build New
Record
Control Key
Copy Record
onto
InterrrEdiate
File
End'
of
Record
Branch to
CI-Eck R ec or
More
Records
Branch to
Return
0
Branch to
Check Record
Sort
on New
Control
Key
Approved For Release 2000/05/08: JAiRbP78-03952A000100070001-5
SE Kti
Approved For Release 2000/05/08SEWORDP78-03952A000100070001-5
Figure 7-9
Merge R
Examine All
Records for
Common
Control Key
Matc h & IVkrge
Records with
Comann Key
into a New
Co mbinscl Record.
Copy New
Combined
Record onto
Output File
Copy Uncorr
bined Records
onto Output
File
More
Records
Branch
to
Return
CA)
)
Branch to
Examine
-86-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
MI
ail
Approved For Release 2000/05/6?Ptifet-RDP78-03952A000100070001-5
record is located in a system generated work area that was
created for this purpose by the EXTRACT command. No
explicit reference need be made to the work area by the
user. Its presence is assumed.
The data elements that may be specified include terms,
tally results, accumulation results, computational results
and defined terms. The command operates only upon record
elements that have been previously extracted.
If hit processing is under FFT control (see
Appendix 7.A.3. for a complete description of FFT Control),
the data being supplied and the action of appending the
data must obey the rules and conventions established by
the File Format Table of the specified input file. Failure
to do so will result in error conditions. The fundamental
concern of FFT control is to insure that files produced by
hit processing commands are still defined by a File Format
Table. Thus enabling these files to be subsequently
processed by the CHIVE system programs.
- Processing Logic: See Figure 7-10 APPEND Logic
7.8.3.2. DELETE
COMMAND LIMIT SEARCH CRITERIA FILE DATA
Record Phrase Term I 0
(Ident)
DELETE (Number) (Ident) (Ident)
(Ident) (Ident) (Ident)
HIT PROCESSING
-87- Editing Commands
Approved For Release 2000/05/085.E9VDP78-03952A00010007J001-5
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Figure 7-10
Append
Branch
to
Delete
Store
Return
Load All
Command
Operands
Get File
Format
Table
Append Data
o Record in
Extract
Work Area
Branch to
Error
Routine
Approved For Release',1000/05/08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05*PTIA-RDP78-03952A000100070001-5
Figure 7-10
Append
Branch
to
Write
Routine
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
SECRET
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Command Description: DELETE causes the terms,
phrases, or records specified in the search criteria to
be deleted from the record located in the work area. As
in the APPEND command, the record elements being operated
upon were introduced into the work area by the EXTRACT
command.
If FFT control is specified in the job specification,
the delete operation must obey all rules established by
the File Format Table of the input file. Failure to
do so will result in an error condition.
- Processing Logic: See Figure 7-11 DELETE Logic
7.8.3.3. REPLACE
COMMAND
LIMIT
SEARCH CRITERIA FILE DATA
RECORD PHRASE TERM I 0
REPLACE (Number) (Ident) (Ident)
(Ident) (Ident)' (Ident)
- - (Data)
Command Description: REPLACE causes the data
elements specified in the DATA field to be substituted
for the terms or phrases specified in the search criteria,
and the result located in a system generated work area. As
in APPEND and DELETE, the record elements being operated
upon were placed inthemork area by the EXTRACT command.
- Processing Logic: See Figure 7-12 REPLACE Logic
-90-
HIT PROCESSING
Editing Commands
7.8.3.3.
Approved For Release 200SeM18 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/0546UIA-RDP78-03952A000100070001-5
Figure 7-1)
Delete
Store
Return
Load All
Command
Operands
Branch
to
Replace
Get File
Control
Table
Delete the
Specified
Record
Elements
Branch to
Error
Routine
V
-al--
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08SECIMRDP78-03952A000100070001-5
Figure 7-11
Delete
Increment
Delete
Counter
Branch
to
Write
Routine
-92-
Branch
to
Return
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
SECRET
Approved For Release 2000-A5/083:2CIA-RDP78-03952A000100070001-5
Replace
Branch
to
Merge P
Get File
Control
Table
Branch to
Error
Routine
Branch to
Error
Routine
ReFlace Record
Element with
Data
Specified
Approved For Release 2000/414-RDP78-03952A000100070001-5
Approved For Release 2000/05/08SOCR-RDP78-03952A000100070001-5
Figure 7-32
Replace
Increment
Replace
Counter
Branch
to
Return
Branch
to
Write
Routine
4 -
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/6ficitTA-RDP78-03952A000100070001-5
7.8.3.4. DEFINE n
COMMAND LIMIT
SEARCH CRITERIA
FILE DATA
Record Phrase Term I 0
DEFINE n
(Ident) (Ident)
(Ident) (Ident) (Ident) (Name)
DEFINE n uniquely identifies a term or phrase in a
record according to the specified search criteria. The
DEFINE command is used in conjunction with other hit
processing commands. For example, when used in conjunction
with the arithmetic commands it allows the user to uniquely
"define" the terms that enter into a computation. In
addition, it insures control via test conditions that the
required value will not occur more than once in the
argument.
7.8.4. THE ARITHMETIC COMMANDS
7.8.4.1. TALLY n
COMMAND LIMIT
SEARCH CRITERIA FILE DATA
RECORD PHRASE TERM I 0
jdent)
TALLY n number (Ident) (Ident)(Ident)(Name) -
(Ident) (Ident)(Ident)
HIT PROCESSING
Arithmetic Commands
7.8.4.1.
-95-
Approved For Release 2000/05/08s:EekiiRDP78-03952A000100070001-5
Approved For Release 206?Ng68 : CIA-RDP78-03952A000100070001-5
Command Description: TALLY n causes a numbered tally
counter (N) to be incremented by one each time that the
specified search criteria are satisfied, for either a
Record, Phrase, or Term specification. It operates upon
the named Input File and stores the result in a symbolic
location named TALLY n. This result may be referenced by
name and its value used as an operand by any command group
including the other arithmetic commands.
The tally counter is reset to zero for each new
record brought into the job when Phrase or Term search
criteria are specified. However, specification of Tally
Records provides a running count of all "hit" records and
the tally counter is not reset until the end of the
retrieval transaction.
Optional inclusion of a value in the Limit field
will cause the tally operation to cease when the count
equals the specified value.
- Processing Logic: See Figure 7-13 TALLY n Logic
HIT PROCESSING
Arithmetic Commands
7.8.4.1.
-96-
Approved For Release 2liCEI/W.108 : CIA-RDP78-03952A000100070001-5
Approved For Relealgul0D0105/1*%?X-RDP78-03952A000100070001-5
Tally
Tally
Garman:1
Branch to
Accum (N)
_
V
Store
Return
Define the
Counter
Tally (N)
Branch to
Error
Routine
Approved For Release 20000578V: CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0?E:CM-RDP78-03952A000100070001-5
Figure 7-13
Tally
Reset
Counter
Tally (N)
Add One to
Counter for
Each Record
Phrase or Term
Branch
to
Check
Record
(1)
-111
Branch
to
Return
-98-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/SKRUA-RDP78-03952A000100070001-5
7.8.4.2. ACCUMULATE
COMMAND LIMIT SEARCH CRITERIA FILE DATA
Record Phrase Term I 0
ACCUM n (Number) (Ident) (Ident) (Ident)(Name) -
Command Description: ACCUM n causes the term
operand(s) specified in the Search Criteria to be
arithemtically summed, and the results are stored in a
symbolic location named ACCUM n. The command operates
upon the named input file. If a value is inserted in the
Limit field, the accumulation operation will cease when
the number of operations performed reaches the specified
Limit value.
All search criteria term operands must be numeric.
Detection of a non-numeric character in any term operand
will create an error condition.
- Processing Logic: See Figure 7-14 ACCUM n Logic
HIT PROCESSING
Arithmetic Commands
7.8.4.2.
-99-
Approved For Release 2000/05/08: CIA:RDP78-03952A000100070001-5
SECRET
Approved For #6168se-210t0/05/08 : CIA-RDP78-03952A000100070001-5
Accumulate
KINAccumN
Comman
?
Store
Return
Define the
Accumulator
Accum (N)
Load All
Command
Operands
1 Branch to
o'd Compute (N)
File
Name
Equal
Check
Rec ord.
earc
Criteria
Equal
MP.
Branch to
Error
Routine
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Relpqm2600p/9?/08 : CIA-RDP78-03952A000100070001-5
Accumulate
erm
Numeric
Reset
Accumulator
Accum (N)
Add Hit
Terms to
Accumulator
More
Records
Branch to \
Error
Routine
/
Increment
Accum
Counter
Branch
to to
Return
Branch
to
Check
Record
0
Approved For Release 29 L05/0?Ecctot-RDP78-03952A000100070001-5
Approved For Release 20985708 : CIA-RDP78-03952A000100070001-5
7.8.4.3. COMPUTE
COMMAND LIMIT SEARCH CRITERIA FILE DATA
Record Phrase Term I 0
COMPUTE n (Number)
- (Name)
(Expression)
Command Description: COMPUTE n causes the computational
expression inserted in the Data field to be executed and
the result stored in the symbolic location named COMPUTE n.
This expression is stated in the following form:
Data Field
Operand OP Operand . . . OP Operand
Where "OP" can be any of the following:
* Multiply
/ Divide
+ Add
- Subtract
Operands may be either numeric valued tags, TALLY or
ACCUM results, or numeric constants. The symbolic location
Compute n is reset to zero for every record encountered in
the job.
- Processing Logic: See Figure 7-15 COMPUTE n Logic
HIT PROCESSING
Arithmetic Commands
7.8.4.3.
-102-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Relerg.ppoiak/966.4K-RDP78-03952A000100070001-5
Compute
Store
Return
Define the
Accurrulator
Compute (N)
cck.11.ame
N
qual
Branch
to
Return
Branch to
Error
Routine
Branch to)
Extract
Approved For Release 2004105/016?C1A-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0?MAIRDP78-03952A000100070001-5
Figure 7-15
Compute
4 , ,
Operands
Equal
?
Extract
Operands
from
Ex ression
Stack
Operands &
Operators
Branch to
Error
Routine
N
ra'NN
AC
Result
Reset
Accumulator
Compute (N)
Approved For Re19), 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/03Kalk-RDP78-03952A000100070001-5
Figure 7-15
I Compute
N>S.
(noi
Execute
Multiplys
Store Re sults
in
Compute (N)
Execute
Divides
Execute
Adds
Store Re sults
in
Compute (N)
rj
Store Results
in
Compute (N)
Execute
Subtracts
Increment
Compute
Counter
Store Results
in
Compute (N)
Branch
to
Return
Approved For Release 2688/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08sPERTRDP78-03952A000100070001-5
Figure 7-15
Compute
Branch
to
Check
Record
-106-
Branch
to
Return)
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/138CRERA-RDP78-03952A000100070001-5
7.8.5. WRITING COMMANDS
7.8.5.1. WRITE
COMMAND LIMIT SEARCH CRITERIA FILE DATA
Record Phrase Term Input Output
WRI TEP
WRITER (Number) (Name)
Command Description: WRITE occurs in two modes, WRITE?
and WRITER. However, the form of the commands and the
structure of the operands is identical. Basically, WRITE
causes the record elements contained in the EXTRACT work
area to be written onto an output file whose name is specified
by the Output File field. When WRITER is specified, the
entire contents of the work area are written out as a
single record. When WRITEP is specified, each phrase in the
work mrea is written out as an individual record.
The WRITE command operates upon the contents of the
work area only and acts as the exit point of a group of
hit processing commands. It is always paired with at
least one EXTRACT command.
The Limit operand causes the WRITE operation to
terminate when the number of records written equals the
Limit value.
- Processing Logic: See Figure 7-16 WRITE Logic
-107-
HIT PROCESSING
Writing Commands
7.8.5.1.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Figure 7-16
Write
Branch
to
Copy
Load All
Cornmand
Operands
Set Up
Output
File
Write Record
in Work Area
as Single
Record
Write Each
Phrase in W
as Separate
Record
Approved For Release 200{3M/08 : CIA-RDP78-03952A000100070001-5
_SECRET
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Figure 7-16
Write
Branch
to
Extract
-109-
Returnto
Branch
Approved For Release 2000/05/0Aep-RDP78-03952A000100070001-5
Approved For Release 2iNt3508 : CIA-RDP78-03952A000100070001-5
7.8.5.2. COPY
COMMAND LIMIT
SEARCH CRITERIA FILE DATA
Record Phrase Term I 0
COPY (Number) (Ident) - -
(Ident) (Ident) - (Name)(Name)
(Ident) (Ident) (Ident)
Command Description: COPY causes the record elements
specified in the Search Criteria field to be written onto
an output file whose name is specified by the Output File
operand.
The command operates upon records, terms, or phrases,
contained in the named input file. If a value is inserted
in the Limit field, the COPY operation will terminate when
the number of records copied equals the specified Limit.
When the COPY command is used, there can be no additional
manipulation of the record elements specified.
- Processing Logic: See Figure 7-17 COPY Logic
7.8.5.3. EXPLODE
COMMAND LIMIT
SEARCH CRITERIA FILE DATA
Record Phrase Term I 0
EXPLODE (Number) (Ident) (Ident) - (Name) (Name) -
(Ident) (Ident) (Ident)
HIT PROCESSING
Writing Commands
7.8.5.3.
-110-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved FotESEAse72000/05/06KM-RDP78-03952A000100070001-5
Store
Return
Load All
Command
Operands
Branch
1to
Explode
..
)
ipiBranch to
Error
Routine
Check
A
Record
-111-
Approved For Release 2000/05/08: CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08SWRDP78-03952A000100070001-5
Figure 7-17
Copy
Write Hit
Element alb
Output File
-112-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08EMFA-RDP78-03952A000100070001-5
Command Description: EXPLODE causes the generation
of a separate, distinct record for each term or phrase
specified in the Search Criteria. It operates upon the
input file so designated and writes the resultant records
onto an output file whose name is specified by the Output
File Operand.
EXPLODE is a variant of COPY. There is one important
difference, it allows terms and phrases to become separate
records. As a result these new files can be regrouped for
meaningful displays. It should be noted that the use of
EXPLODE causes the record relationships to be lost.
- Processing Logic: See Figure 7-18 EXPLODE Logic
7.8.5.4. PRINT
COMMAND LIMIT
PRINT (Number)
SEARCH CRITERIA
Record Phrase Term
FILE
0
DATA
-(Name)(Name)(Parameters)
The PRINT command causes the Input File records to
be routed to and processed by the report programs. Sort
sequence, format, and other control parameters are indicated
in the Data field.
-Processing Logic: See Figure 7-19 PRINT Logic
HIT PROCESSING
Writing Commands
7.8.5.4.
-113-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Figure 7-18
Explode
Store
Return
Set Up
Output
File
Load All
Command
Operands
Branch
to
Return
Branch
to Error
Routine
Approved For Releas 2000105/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Figure 7-18
Explode
Write Hit
Element onto
Output File
Branch
to
Check
Record
-115-
Increment
Explode
Counter
Branch
to
Return
Approved For Release 2000/05/08~DP78-03952A000100070001-5
Approved For Release 2000/05/Ifecta-RDP78-03952A000100070001-5
Figure 7-19
Print
Store
Return
Force Out
All Job Find
Counters
Clear All
Flit Processing
Work Areas
Branch
to
Output
Routine
oilBranch to
Error
Routine
Approved For Release 2d0905/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/88CR1PA-RDP78-03952A000100070001-5
Chapter 7.9.
INPUT PROCESSING
7.9.1. TYPES OF INPUT
CHIVE EDP inputs will consist primarily of data and
commands (maintenance and retrieval) entered on coding
sheets. These will be processed by the optical page
reader. In the initial system structuring information
will be entered via assembled or compiled tables.
Examples of structuring inputs are File Format Tables
and report format definitions.
Control statements will also be directed at both
Operating System/360 and the operational programs.
Control information (such as job start, job end, and job
priority) are directed at the operating system. More
detailed information such as input data file identifications
are used by operational programs.
7.9.2. INPUT LANGUAGE PROCESSOR
The input processor or generator interprets and
analyzes input command statements and generates, as output,
a list of references (program calls) to problem programs.
INPUT PROCESSING
Language Processor
7.9.2.
-117-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 200610316E : CIA-RDP78-03952A000100070001-5
7.9.2.1. Placement
Because of the expected use of Operating System/360
in the CHIVE system, a logical placement may be within
the 05/360 input processing framework. Thus, two
alternatives present themselves--either place the
generator within OS/360, or outside of OS/360 as the
first problem program to handle the input stream.
Tentative indications are that the generator may be
best used as the first operational program.
A Transaction Control Block (TCB) will be set up
for each job being compiled. Core storage blocks will
be allocated to the TCB, and the generator output will be
stored in these blocks. The final action of the input
generator can be a transfer to the first generated
instruction. The instructions will consist primarily
of LINK macros which will cause transfers to the program(s)
required to execute each command.
7.9.2.2. Design
The various points concerning the design of the
generator can be narrowed to two subjects--complexity
of design and the method of generation.
The processing complexity of a generator is directly
proportional to how similar the structure of the user
INPUT PROCESSING
Language Processor
7.9.2.2.
-118-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/6{PtTA-RDID78-03952A000100070001-5
language is to the problem program system. A close
similarity exists when one statement (command) in the
language can be processed by one problem program. In
this case, the generator would only need to perform a
single table look-up operation to translate each
statement into a program "call." In contrast to this
simple approach, the generator may be required to
perform a complicated analysis on each statement in
a sophisticate user language. The CHIVE language
is simple enough to support the first approach.
The generator can perform its generation functions
in two ways--(a) complete preprocessing, or (b) interpretive.
Complete preprocessing is defined as a thorough analysis
of the entire set of commands before any of them is
executed. When the preprocessing is completed, the commands
are in a form usable by the programs; and no further pass
is needL:.. through the generator. On the other hand an
interpretive generator decodes one statment at a
time, executes the statement, then steps to the next
statement. The process is continued until the job is
completed.
7.9.3. COMMAND MACROS AND STANDING JOB DEFINITIONS
In addition to the regular jobs containing
explicit command, transactions can contain references
-119-
INPUT PROCESSING
Job Definitions
7.9.3.
Approved For Release 2000/05/08 Ri()ERDP78-03952A000100070001-5
Approved For Release 2009MOIS : CIA-RDP78-03952A000100070001-5
to commands that were previously defined to the
generator in the following two ways:
- Command Macro Definition Statements
- Standing Job Definitions Statements
These definitions can be used in a load-and-go job
or simply as a stand alone "definition job."
Figure 7-20 shows the input command processing performed
on jobs containing the various types of command statements.
There are major differences between the two types
of definition statments. A command macro is used to
build a command character string, which may contain
parameters. It is designed to simplify coding of
repetitive command strings and would normally be used
within a job. On the other hand, a standing job
definition "defines" a complete job which may include
control statements, command statements, and command
macros.
7.9.3.1. Command Macros
A command macro definition specifies a set of
contiguous standard CHIVE commands and a set of dummy
parameters which will be "filled in" when this macro is
called by a macro statement. Once defined and entered
into the system, any subsequent job can use this macro
INPUT PROCESSING
Job Definitions
7.9.3.1.
-120-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
IL Lit
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Figure 7-20
Initial
Command Macro
Definitions
Pre -
Scanner
INPUT COMMAND STATEMENTS
Regular*
Command
Job
Pre-
Scanner
A
V
Standing
Job
Library
Generator **
Execute Job
Initial
Standing Job
Definitions
Pre-
Scanner
_
Generator
*May include command macro state-
**Standing jobs bypass this stage because
ments or e a standing job request,
they have been processed previously by
b
Approved For Release 2000/05/08 : CIA-RDP78-03962A000400070001-5
cn
rn
r)
73
rn
-4
Approved For Release 2000WeiT. CIA-RDP78-03952A000100070001-5
by referring to it by name. It will be "compiled" each
time that it is used in a job. When a command macro
is detected by the pre-scanner section of the generator,
the following events occur;
- Locate the appropriate command macro skeleton
in the macro library.
- Fill in the dummy parameters with the real ones
given as part of the macro.
- Replace the command macro with the included
standard CHIVE commands.
7.9.3.2. Standing Jobs
The creation of a standing job involves
a complete
specification of a entire job (some parameters may be
specified at execution time). The job is processed
through the generator as a normal job, but instead of
being executed, it is given an identification tag and
stored in the standing job library. Henceforth, this
job need only be requested by identification number
(and, optionally, some parameters supplied to it), and
it will be pulled from the library and executed. It
is "compiled" once, but may be used as often as
necessary.
A job, if it is not a standing job and contains no
references to command macros, is processed directly by
the generator. A job that refers to one or more command
macros undergoes some additional pre-scanner steps
INPUT PROCESSING
Job Definitions
-122- 7.9.3.2.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0Vqtk-RDP78-03952A000100070001-5
which merge the requested commands into the job. A
request for a standing job is interpreted by the pre-
scanner which starts the standing job through the system.
7.9.4. INPUT JOB BATCHING
Transactions that enter the CHIVE System can be
introduced singly or grouped together into a batch job.
A job composed of a single transaction poses no require-
ment for special processing, for it is immediately
handled by the generator and executed. Batch jobs, however,
require several additional processing steps.
7.9.4.1. Definition of a Batch
A batch job is defined as that set of CHIVE
transactions that are present between Operating System/360
"job header" and "end" control statements. The job
control capabilities of the OS/360 are consequently used
to conL?A batches of CHIVE transactions. Batching in the
CHIVE system, however, requires two levels of control
statements. Between the two OS/360 control cards are
CHIVE control statements that are necessary to define
completely the individual transactions within the batch.
These CHIVE control statements will be interpreted first
by the CHIVE generator and subsequently by the CHIVE
problem programs. Transactions using different files
INPUT PROCESSING
Job Batching
7.9.4.1.
-123-
Approved For Release 2000/05/08s.E8VDP78-03952A000100070001-5
Approved For Release 200NOWN : CIA-RDP78-03952A000100070001-5
and having different priorities may be included in the
same batch.
7.9.4.2. Batch Processing
The initial processing of batch jobs involves a
series of preprocessing steps that are performed by
Phase I and II of the CHIVE generator. These steps,
some of which may be performed asynchronously, are
discussed below. Figure 7-21 shows the major
interactions between the various steps.
Phase I - Sorting, Compiling, and Separation: All
the input transactions within a batch will be sorted and
classified into three groups -
- Retrieval and conditional maintenance transactions
- Unconditional maintenance transactions
- Special page reader inputs, e.g., document index
header and content phrase sheets which comprise
a new record
Phase I - Analysis and Generation: The test-lines
and commands of each transaction will be compiled into a
sequence of program calls.
Phase II - Separation: All transactions that
reference direct access files are ready for processing and
are introduced immediately into the system. Transactions
INPUT PROCESSING
Job Batching
7.9.4.2.
-124-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2006i65/687:-611A-RDP78-03952A000100070001-5
CHIVE INPUT PROCESSING
Optical
Page
Reader
Magnet ii.
Tape
nard
Reader
OS/360
Staging
Device
Sorting
Separation
Analysis
Generation
Unconditional
Maintenance
Transaction.,://
(cetrieval &
?......
Conditional
Maintenance
Transactions
Special Page
Reader Inputs
Transactions
Separation
"Tape" File
Transactions
Random
Access
Transactions
Tape File
Sub-batching
Approved For Release 2000L dttfetTA-R P78-03952A000100070001-5
-125-
Approved For Release 2000/0?TagbIA-RDP78-03952A000100070001-5
that process a serially access tape file receive
additional processing.
Phase II - Tape File Sub-batching: To optimize the
accessing of magnetic tape files Phase II will analyze
the compiled transactions and rearrange them into
meaningful (sub) batches associated with each individual
tape file. After this is done, the batches are ready
for execution.
7.9.5. INPUT FUNCTIONS
As a summary, a list of the major input functions
are given below. All of them are directed at preparing
a job for execution in the CHIVE system.
- Read input (data, commands, and control informa-
tion) from the system input devices.
- Perform checks on input information (such as,
formatting and seqp,ftbibng checks).
- Decompose command statements
(Phase I of the Generator).
into program calls
Partition batch jobs and set up optimum processing
steps (Phase II of the Generator).
- Provide input services (such as command macro and
standing job capability).
INPUT PROCESSING
Input FOnctions
7.9.5.
-126-
Approved For Release 2000/0WR:E9IA-RDP78-03952A000100070001-5
Approved For Release 2000/05/0%EcUA-RDP78-03952A000100070001-5
Chapter 7.10
OUTPUT PROCESSING
7.10.1. OUTPUT ROUTINE DEFINITION
Output reports will provide the vehicle for conveying
to the information analyst or a research analyst the
data extracted from the CHIVE files. Based upon their
intended use, the urgency of the response and the files
concerned, the makeup of a printed page may vary from a
direct presentation of the file contents, with minimum
editing, to a well-organized composition of selected
items, arranged hierarchically to accentuate the most
significant fields of information, with the more
discriminating fields reordered so as to simplify their
interpretation. Summary features of complete reports may
be fur.L'= isolated from the body of the report, by
placing them on cover pages, to facilitate a preliminary
assessment of its contents.
The initial routines for report generation will be
prepared to fit the general requirements for the most
common reports anticipated. They will be maintained in the
system library subject to call by the analyst, with a
OUTPUT PROCESSING
Routine Definition
7.10.1.
-127-
Approved For Release 2000/05/08 : CIARDP78-03952A000100070001-5
SECRET
Approved For Release 2@N4108 : CIA-RDP78-03952A000100070001-5
PRINT command. Each report can have parameters
which must be specified in the PRINT command. An
enumeration of the reports to be defined for the various
files is included in Table 7.C-1 Same Report Formats
and a number of sample layouts for such reports appear
in Appendix 7.C.
A report will contain three types of pages as
shown in Figure 7-22: cover pages for the complete
report, boundary pages for the body of the report, and
content pages which represent the bulk of the report.
Variation in the appearance of content pages will
be due to differences in the data contained in individual
records being represented or due to changes in pages, such
as the page number. The format will generally be fixed,
subject to any conditions placed upon lines or parts of
lines in the report specifications.
Variation in the appearance of boundary pages will
be due primarily to the report itself, though these
pages may actually contain some information extracted
from records. They will contain enough information about
the report to assure that if the cover pages are removed
the nature and basis of the report can be determined.
The first, or initial, pages will contain the identity
OUTPUT PROCESSING
Routine Definition
7.10.1.
-128-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
SECRET
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Figure 7-22. PAGE ARRANGEMENT - CHIVE REPORTS
-129-
Approved For Release 2000/05/0%iederRDP78-03952A000100070001-5
Approved For Release 2000geM: CIA-RDP78-03952A000100070001-5
of the report and the request which caused it to be
prepared. The last, or final, pages will contain
summary information such as totals or counts for
the complete output file.
Variations in the appearance of cover pages will be
due to the entire retrieval transaction which issues
the PRINT command. Front covers will contain clear
identification of the request and the requestor with
such routing information as is available. It may also
present any data descriptive of the file which is known
before the body of the report is printed. Back covers
will similarly contain identification and routing
information and will include entries derived from the
entire file reported upon, such as the identity of
records (or documents) from which the reported results
were extracted. This may be used as a request sheet for
subsequent retrieval of selected documents after
appropriate annotation by the analyst.
7.10.2. REPORT-RELATED PROCESSING FUNCTIONS
In order that information published in reports be
as meaningful as possible to the reader the data may be
rearranged and translated from its compact internal form.
These processes will be perfomed automatically as part
OUTPUT PROCESSING
Processing Functions
7.10.2.
-130-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0PW-RDP78-03952A000100070001-5
of the report generation routines on the basis of
parameters supplied by the analyst in calling for the
report with a PRINT command. The routines which complete
these processes may operate at several levels during
the report preparation.
At the level of the transaction, the specifications
for the transaction may be examined and enough data
extracted and stored in cover and boundary page print
areas to define the request number, the name and location
of the requestor, and, the search criteria. Where no
specific PRINT command appears in the transaction
specifications, but printing is implied by special commands
such as TALLY or ACCUM commands, the entire report generation
will be accomplished at the transaction level.
At the level of the output file the process is
one of sorting. The PRINT command in the retrieval
transaction specifications which initiates the report
generation may designate one or more terms which will
serve as keys for sorting. This will cause the records
in the file to be sorted in the order indicated prior to
the actual generation of the report.
At the level of the individual record is the
process of cross-referencing source citations. As output
OUTPUT PROCESSING
Processing Functions
7.10.2.
-131-
Approved For Release 2000/05/08iaTIRDP78-03952A000100070001-5
Approved For Release 200N1NG$ : CIA-RDP78-03952A000100070001-5
records are built up the identity of the document
from which the information is derived may be attached
to the item placed in the record. Each item and its
source document identity may both be linked with a
unique sequential flag so that the two parts may be
separated and later reassociated (for example, see
Figure 7.C-7 in Appendix 7.C.). This will permit
the printing of items of information and cross referenced
source citations during report generation.
At the level of the term the report-related processing
functions are based upon the desire to decode or translate
the tags and/or values of the terms and to simplify the
representation or interpretation of the meaning of these
parts.
The decoding of tags will transform the internal repre-
sentation of the tag into either a fixed or a variable
length equivalent by means of a table lookup. If the
ultimate user is familiar with the 3-character mnemonic,
this form may be used; otherwise a narrative equivalent
will be used. The equivalents must be inserted into the
translation table at the time the decoding routines are
prepared.
Document titles may be converted into a more easily
interpreted form by permuting the contents cyclically,
OUTPUT PROCESSING
Processing Functions
7.10.2.
-132-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0SECOMPRDP78-03952A000100070001-5
to align key words in a fixed position and replicating
the record for each such key word. The copies may be
sorted on the key words in a single group for the record
copied or all copies of all records may be assembled as
a single file and sorted on the key words which appear at
the designated position.
7.10.3. GENERAL REPORT GENERATING FUNCTIONS
A very comprehensive choice of report generation
capabilities can be provided with the combination of:
- A varied selection of line types
- Conditional control over line or line element
printing
- Line or column overflow for variable length fields
- Vertical spacing through channel carriage control
- Editing and decoding by table lookup or subroutine
- Sorting of a file prior to report generation
- Arithmetic actions taken at report time
7.10.3.1. Line Types
Line types fall into two groups, those which
contain only non-record data which is either invariant
or varies only with respect to the transaction or the
progress of the report itself, and those which contain
OUTPUT PROCESSING
Generating Functions
7.10.3.1.
-133-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 20Q@ONP8 : CIA-RDP78-03952A000100070001-5
some record data which may vary between records of the
file being reported.
Lines which contain only non-record data may be
composed from actual character strings supplied at
definition time, character strings previously designated
for use in other line definitions or elements of data
derived from the transaction definition statements.
Figure 7-23 illustrates the following non-record line
types:
- Leader lines
Trailer lines
- Title lines
which are to be printed at the
top and bottom, respectively,
of each page printed
which may appear on the covers
or initial pages or on other
pages conditionally
Lines which contain record data may be composed
from actual character strings supplied at definition
time, character strings previously designated for use
in other line definitions, elements of data derived from
the transaction definition statements, data derived from
terms extracted from data records which may have been
edited or decoded, or data obtained as the result of
arithmetic actions. The record line types are:
- Label lines which may appear on initial,
content or final pages,
OUTPUT PROCESSING
Generating Functions
7.10.3.1.
-134-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
25X1A
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Approved For Release 200NRCI8 : CIA-RDP78-03952A000100070001-5
- Term lines
- Phrase lines
which may appear on content
pages or the content portion
of initial or final pages.
In term lines the terms will
be assigned to fixed column
locations within a line on
the basis of their tags.
Overflow, if permitted, will
remain within the column
which may appear on content
pages or the content portion
of initial and final pages.
In phrase lines the terms will
be positioned on a line in the
order in which they appear in
the phrase. Possible overflow
conventions appear in Section
7.10.3.3., Overflow, below
- Final lines which usually appear conditionally
on the terminating page.
The likelihood of occurrence of a line of a particular
type in a specified type of page is shown by the following
matrix in which
R indicates the line type is required
P indicates the line type is probable
O indicates the line type is optionally possible
Page Type Front Initial Content Final Back
Line Type Cover . Page Page Page Cover
Leader R R R R R
Title P P 0 0 P
Label 0 0 0
Phrase 0 P 0
Term 0 P 0
Final P 0
Trailer R R R R R
OUTPUT PROCESSING
Generating Functions
7.10.3.1.
-136-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/actfIA-RDP78-03952A000100070001-5
Briefly the line samples in Figure 7-23 may be described
as follows:
- Leader Line: This consists of date, security
classification and page number. It will be
printed at the top of each page of the report.
- Title Lines: These consist of a title for this
type of transaction and information about the
request which will be extracted from the
transaction specifications.
- Label Line: This is a line of headings
which may appear above columns of information
describing documents selected during the
retrieval process.
- Phrase Lines and Term Lines: These are examples
of the presentation of one record of information
which has been selected in response to a request
concerning men attending a particular conference.
In the case of the phrase lines the fields are
printed successively across the page with a fixed
3-character space between fields and the actual
phrase content printed in its entirety.
In the term lines the same fields are placed in
columns, 30 characters wide, of which 3 characters
are used for the tag, 1 character (-) is used to
separate the tag and value and 26 characters are
used for the value, the remaining characters of
the value being lost due to truncation.
In both these line samples it is assumed that the
requestor wishes the tags to be printed as
3-character mnemonics, rather than having them
decoded into natural English.
- Final Lines: These are examples of information
which may be presented to summarize the results
of a search through the document index records.
- Trailer Line: This contains information extracted
from the transaction specifications as well as the
security classification. It will appear at the
bottom of each page printed.
OUTPUT PROCESSING
Generating Functions
7.10.3.1.
-137-
Approved For Release 2000/05/08SEcRt] : Cl&RDP78-03952A000100070001-5
Approved For Release 2002008 : CIA-RDP78-03952A000100070001-5
7.10.3.2. Conditional Control of Print Lines
In addition to the conditional retrieval of
records, it is possible to condition the printing of
reports made from these records. Tests of whether a
field is greater than, less than, equal to or not equal
to specified values may be accomplished by test conditions
similar to those employed in retrieval and conditional
maintenance. The serial character of report generation
and the ordering of records in files or data in records
introduce additional conditions which can be exploited
for the enhancement of output. Particular conditions
which may be applied to report generation are:
- Change or repetition of the value of a
which recurs in successive print lines
being reported.
term
or records
- The completion (exhaustion) of reporting of an
identifiable subset of an ordered set of values
or data collections, at any level of the hierarchy.
- The satisfaction of test conditions by record data.
- The satisfaction of test conditions by non-record
data.
Examples of the use of such conditions might be:
- To cause a label line to be printed whenever a
record changes.
- To suppress the printing of part of a line
which contains terms which are designated as the
sort key whenever that part would not change from
line to line.
OUTPUT PROCESSING
Generating Functions
7.10.3.2.
-138--
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/CMCM-RDP78-03952A000100070001-5
- To cause the printing of special intermediate lines
when all records in the file containing a parti-
ular value of a sort key term have been processed.
- To cause a particular line format to be used for
records meeting test conditions.
- To cause printing to be halted after a specified
number of records has been processed.
The manner in which these tests may be applied to
the various line types is:
- Leader, title, final and trailer lines or line
elements may be conditioned by tests on non-record
data.
- Label or term lines or line elements and
phrase lines may be conditioned upon tests,
changes or completion of record data.
- Label and term lines may be associated with
one another so that the label line will appear
only if its associated term line does. The
conditional control of label lines is therefore
subordinated to the control of the term line
associated with it.
7.10.3.3. Overflow
The printing of term lines may cause overflow
in two ways, (a) the variable number of occurrences of
a tag (i.e., different values for the same tag) or
(b) the variable length of values of terms with a given
tag which is to be printed in a fixed-length field. The
first cause, variable occurrences of the same tag, may
be accommodated by printing the term lines as many
times as the greatest number of occurrences of any term
OUTPUT PROCESSING
Generating Functions
7.10.3.3.
-139-
Approved For Release 2000/05/08: CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2001MSKRIT : CIA-RDP78-03952A000100070001-5
type which appears in the term line. For those term
types having fewer values than this greatest number,
blank fields will appear for each term line element
for that tag after its values have been exhausted. As
a result the appearance of a specified term line type may
be:
Tag 1
Value 1
Value 2
Tag 2
Value 1 with
Overflow
Value 2
Value 3
Value 4
Tag 3
Value 1
Value 2 with
Overflow
For the second case, variable length values, there
are three options which may be selected in cases where
the term value is not of constant length:
- Values may be converted to a fixed length by use
of a predefined equivalence table.
- Overflowing values may be truncated to the length
available with attendant loss of the excessive
characters
- Overflowing values may be carried over to
succeeding lines in that column with some editing
to preserve readability. This is the procedure
assumed in the example above.
The printing of phrase lines, because of the greater
variability possible in phrase contents, offers more
-140-
OUTPUT PROCESSING
Generating Functions
7.10.3.3.
Approved For Release 200RM8 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/08E:CQUE-RDP78-03952A000100070001-5
alternatives in format and the concomitant handling of
overflow:
- The terms of a phrase may be sequenced horizontally
across the page or they may be arranged vertically
in a column only one term wide.
- A term from the phrase may be arranged in a single
line as tag followed by value, or in a double
line as a tag over a value.
- Columns for terms may be in fixed positions or may
float in response to the maximum width of a value
for terms in a phrase.
The nature of the record printout, in phrase-line
format, which will be created by each of the eight
possible combinations of these options are shown in
Figures 7-24 through 7-31.
7.10.3.4. Carriage Control
The format of a report can be affected significantly
by the control of spacing between lines or parts of a
report. This may be attained in three different ways:
spacing a specified number lines, ejecting to the next
page or skipping to a specified line number on the
page. Skipping may cause the carriage to move forward
on the current page or to skip to the line number on the
next page depending upon whether the current line number
before the skip is or is not less than the specified
number.
OUTPUT PROCESSING
Generating Functions
7.10.3.4.
-141-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 20561igg8 : CIA-RDP78-03952A000100070001-5
Figure 7-24
Phrase Print Format 1
Horizontal Arrangement; Single Line; Floating Tera Position
HEALER Tag 1 (xxxxx) Tag 2 (xxxxxxxxxx) Tag 3 (xxxxxxxxxxxxxxx)
Tag 4 (xxxxxxxxxxxxxxxxxxxx) Tag 5 (xxxxxxxxxxxxxxxxxxxxxxxxx)
Phrase 1 Tag 1 (ccxxx) Tag 2 (acxxxxxxxx) Tag 3 (xxxxxxxxxxcxxxx)
Tag 4 (xxxxxxxxxxxxxxxxxxxx) Tag 5 (xxxxxxxxxxxxxxxxxxxxxxxxx)
Tag 6 (
Tag 7 (
' * **** **, * eV ? .-
Phrase 2 Tag 1 xxxxx) Tag 2 xxxxxxxxxx)
-142-
Approved For Release 2gge1jii/08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05411calfk-RDP78-03952A000100070001-5
Figure 7-25
Phrase Print Format 2
Horizontal Arrangement; Single Line; Fixed Term Positi
HEALER Tag 1 (xxxxx)
Tag 3 (
Tag 5 (
xxxxxxx
Tag 7 (xx
Phrase 1 Tag 1 (xxxxx)
Tag 3 (xxxxxxxxxxxxxxx)
Teg 5 (
xxxxxxx
Phrase 2 Tag 1 (xXXXX)
XXX CC
-143-
Tag 2 (xxxxxxxxxx)
Tag 4 (xxxxxxxxxxxxxxxxxx
xx)
Tag 6 (xxxxxxxxxxx
xxxxxxxxxxxx)
Tag 2 C
Tag 4 (xxxxxxx
xx)
Tag 2 xxxxxxicacx)
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 200%figgide : CIA-RDP78-03952A000100070001-5
Figure 7-26
Phrase Print Format 3
Horizontal Arrangement; Dotble Line; Floating Term Position
HEADER Tag 1 Tag 2 Tag 3
(ommtx) (caccxx) (xxxaDucc)
Tag 4 Tag 5 Tag 6
(xxxxxxxx) (raccxxxxx ) (=Dom= )
Tag 7
(accxxxxxxxx)
Phrase 1
Phrase 2
Tag 1 Tag 2 Tag 3
(xxxxx) (loccocx) (xorxxxx)
Tag 4 Tag 5
(xmmaxxxx) (accoommar)
Tag I
(room)
Tag 4
Tag 2
(xmmomc)
Tag 5
(xxxmccxx) (
-14 4 -
Tag 3
(=mom)
Approved For Release 20(40a8 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/08E0M-RDP78-03952A000100070001-5
Figure 7-27
Phrase Print Format 4
Horizontal Arrangement; Double Line; Fixed Term Position
HEADER Tag 1 Tag 2 Tag 3
(xxxxxx) (xxxxxxxx) (cxxxxxxxxx)
Tag 11 Tag 5 Tag 6
xxx
Tag 7
xxxxx
Phrase 1 Tag 1 Tag 2 Tag 3
(10011000) (XXXXXXXX) (cxXxxxxxxx)
Tag 4 Tag 5 Tag 6
XXX
Phrase 2 Tag 1
(cxxxxx) (arcxxxxx)
Ta g 2
-145-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2009akeft : CIA-RDP78-03952A000100070001-5
Figure 7-28
Phrase Print Format 5
Vertical Arrangement; ?1.1:Ea Line; Floating Term Position
Header
Tag 1 (xxxx=a)
Tag 2 (
Tag 3 (coccomoom)
Tag 4 (xxxxx)
Tag 5 (
Phrase 3
Ile
Phrase 1
Tag 1 (
41,11 ?
Tag 2 (xoca)
Tag 3 (moccoccxxx)
Phrase II-
Phrase 2
Tag 1 (xmaroc000loa)
Tag 2 (
Tag 3 (xxxxxx=coocca)
Tag 4 (xxxx)
Tag 5 (=cocoa)
Tag 6 (xxxxxxxxxxxxxxx)
Tag 1 (=ocoaxxxxxxxxxxxxx=ca) Tag 1 (xxxmoopoolax)
Tag 2 (x=000ccooa) Tag 2 (
-146-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08EQM-RDP78-03952A000100070001-5
Figure 7-29
Phrase Print Format 6
Vertical Arrangement; Single Line; Fixed Term Position
Header
Tag 1
Tag 2
Phrase 1
Tag 1 (xcxxxxxxxx)
Phrase 2
Tag 1 (xxxxxxxxxxxx)
Tag 2 ( Tag 2 (xccxxxxx
xxxxx)
Tag 3 (xx Tag 3 (
XXXXXXXXXXXX3=
Phrase 3
Tag 1 (ucxxxxxxxx)
Tag It. (xxxxx)
Phrase it.
Tag 1 (
xxxXx
Tag 2 (xxxxxxxxxxxxxxx Tag 2 (
xxxxxxxxxxxxxx) xx)
Tag 3 (
-147-
.41 *lb'
XXX
Phrase 5
Tag 1 (xxxxxxxxxxxx)
Tag 2 (xxxxxx)
Approved For Release 2000/05/08 : C1A=RDP78-03952A000100070001-5
SEcRt
Approved For Release 20brEgga8 : CIA-RDP78-03952A000100070001-5
Figure 7-30
Phrase Print Format T
Vertical Arrangements; Double Line; Floating Term Position
Header Phrase 1
Tag 1 Tag 1
(cccxxxxxxxxxx) (xxxx)
Tag 2 Tag 2
(cocxxxxxx) (xxxxxxxx)
Tag 3 Tag 3
(ccrxxxx) (cocx)
Tag 4
(xxxxxxxxxxxxxxxxxx)
Phrase 4
Tag 1
* ** o **** *
Tag 2
X-1CXXXXXXX
Phrase 2
Tag 1
Tag 2
(XXX)000000C
Phrase 5
Tag 1
:000 ? ? ?
Tag 2
? ?? *.
Tag 3
lax)
464114P 410 ? 0 411 *0
-14 8 -
Phrase 3
Tag 1
(XXXXXXX)
Tag 2
(-xxxx)
Tag 3
MOD=
Approved For Release 213g19,98 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/actn-RDP78-03952A000100070001-5
Figure 7-31
Phrase Print Format 8
Vertical Arrangement; Double Line; Fixed Term Position
Header
Tag 1
(XLICXXXX
Tag 2
Tag 3
(xxxxxxxxxx=x)
Phrase 1
Tag 1
xxxxxx
Tag 2
(xxxxxxxxx)
Tag 3
xxxx)
Tag 4
C.
-149-
Phrase 2
Tag 1
(moccxxxxx)
Tag
(xxxxxxxxxxxxxxxxxxxxx
Approved For Release 2000/05/08s:EEWERDP78-03952A000100070001-5
Approved For Release 200MET8 : CIA-RDP78-03952A000100070001-5
All three methods of controlling spacing may be
conditioned by test, change or completion conditions
as explained in Section 7.10.3.2. above. They may also be
called for unconditionally, before printing of a
designated line type or between records which are
being reported.
7.10.3.5. Decoding
Decoding may involve the use of subroutines or
tables or a combination of these. The most common
method will be the use of a table in which the argument
will be the internal representation of the item to be
converted and the values will be one or more representa-
tions of the item in a converted from. Where the form
used for reporting is the same as that used for input
there will probably be inverted tables used for input
conversion or encoding.
The increased readability of reports whose fields
are of fixed length tends to encourage the use of
conversion tables. Some of the fields which may lend
themselves to this type of treatment are:
- Tags
- Subject Codes
- Area Codes
OUTPUT PROCESSING
Generating Functions
7.10.3.5.
-150-
Approved For Release 20jeay08 : CIA-RDP78-03952A000100070001-5
?
awl
Approved For Release 2000/05/WM-RDP78-03952A000100070001-5
- Security Classification
- Evaluation Codes
- Dissemination and Collection Codes
- Codewords
Some types of items lend themselves to direct
algorithmic subroutine handling. Dates are more easily
compared internally if they are in numeric form in the
order of year, month and day, whereas for reading the
user is more accustomed to their appearance in the order
of day, month and year. The rearrangement is a trans-
position of characters while the month may be converted
with tables. For similar reasons the characters in the
Intelligence Information Report Number require trans-
position as well as any conversion which may be desired.
7.10.3.6. Sorting
Sorting of a file will be accomplished during
output processing by a system sort program. When any
specifications for a sort key are recognized in the PRINT
command, the entire sort key will be determined. It
will then be specified to the sort program and the sort
will be performed before the file is processed for
printing.
OUTPUT PROCESSING
Generating Functions
7.10.3.6.
-151-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 200Q/0014 CIA-RDP78-03952A000100070001-5
If one or more iterative tags are named in a sort
key there is an implicit exploding of each record on
each sort key value in the record. The exploded records
are printed out in sort order.
7.10.3.7. Arithmetic Functions
Arithmetic functions which may be included as part
of output processing are:
- Tallying of items which satisfy specified test
conditions across records. The tally will be
stored in a six digit field.
- Accumulating sum of numeric fields across records.
The total will be stored in a field twice as
long as the field being accumulated.
- Addition of numeric fields occurring in one
record. The sum will be placed in a field one
digit longer than the longest term in the sum.
- Subtraction of fields occurring in one record.
The difference will be placed in a field one
digit longer than the longest term in the
difference.
- Multiplication of fields occurring in a record.
The product will be placed in a field as long
as the sum of the lengths of the factors.
- Division of one field by another. The quotient
will be stored in a field as long as the dividend.
For arithmetic purposes fields derived from the
performance of arithmetic functions upon fields in
the same record may be considered fields occurring in the
record for further computation.
OUTPUT PROCESSING
Generating Functions
7.10.3.7.
-152-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/ackFIT4-RDP78-03952A000100070001-5
Chapter 7.11.
SECURITY PROCEDURES
7.11.1. PROBLEM STATEMENT
CHIVE EDP security requirements are fundamentally
unique when compared with EDP security requirements
in existing intelligence processing systems. CHIVE
security requirements ultimately stem from the multi-
programming character of the CHIVE system. With
multiprogramming, jobs of different sensitivity are
run concurrently. Multiprogramming gives rise to the
practice of maintaining diverse data files simultaneously
on line--files containing data of different classification.
This is in contrast with the normal mode of operations
in the intelligence community, where jobs are batched
by security classification and run one-at-a-time through
the system. Storage units at one security level are
mounted or dismounted for each batch. Conventionally,
security isolation between files of varying sensitivity
is achieved by physical human intervention, not automati-
cally.
CHIVE, on the other hand, maintains diverse physical
files on-line, ready at all times to serve demands made
SECURITY PROCEDURES
Problem Statement
7.11.1.
-153-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/6KAP. CIA-RDP78-03952A000100070001-5
by the processor. Because the CHIVE data base is all-
source, CHIVE maintains files of data at varying security
levels; but because CHIVE service is tailor-made to
customers of different clearance, output records from
homogeneous files of different classification must be
intermixed, and heterogeneous files must be screened.
7.11.2. MIXED DATA-CLASSIFICATION CONCEPT (ALL SOURCE)
There are two types of CHIVE data files with respect
to security (a) heterogeneous and (b) homogeneous.
Heterogeneous files contain records which are
security controlled. Each record may be at a different
classification level from the others. Homogeneous files
contain records all of which are classed at a single
level, that of the file itself.
One may consider a heterogeneous or homogeneous
logical or physical file, to distinguish the data contents
from the physical storage medium on which they are recorded.
One may also regard a single storage device such as a
disk to consist of one or more physical files (e.g., each
cylinder may be a physical file), or a single storage
device may be one segment of a larger physical file. One
SECURITY PROCEDURES
Nixed Concept
7.11.2.
-154-
Approved For Release 2000igi9kt CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/0?Ealier-RDP78-03952A000100070001-5
logical file may span or comprise one or more or less than
one physical file. Therefore, several logical files may
be stored on one physical file, or one logical file may
be strung over several physical files.
In CHIVE the classification of data must be
distinguished from the classification of the storage
media upon which they reside. All dismountable or fixed
storage units within CHIVE are all source. If a dismount-
able unit is released to an outside agency, the unit is
classified at the level of the most sensitive data it
contains or has contained since last erased for declassifi-
cation of the physical storage medium.*
erased, it is reclassified.
Logical files may be assigned to any CHIVE-internal
physical file. In CHIVE, all allocations of logical
to physical files are allowed, irrespective of the
security levels of the physical or logical files, except
that no logical file may be assigned to a physical file
which is not at a security level at least equal to that
of the data. Of course, since only dismountable devices
releasable to outside agencies are not all-source, any
fixed device may receive any logical file.
If the
unit is
*Note: See Section 7.11.14.4. on protection against laboratory
restoration techniques. A similar technique for dismountable
disks will be developed to permit downgrading of the physical
medium.
-155-
SECURITY PROCEDURES
Mixed Concept
7.11.2.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2006% : CIA-RDP78-03952A000100070001-5
Furthermore, two or more logical files may occupy
the same physical storage device without regard to the
level of the device or to each other, provided that the
level of the device is not exceeded by the level of either
file. Each logical file retains its own security level.
In summary, within the stipulations of the mixed
data-classification concept, any combination of logical
and physical files at any security level may reside
on-line simultaneously. Security isolation of data
is achieved by software checking procedures.
7.11.3. MIXED EDP JOB CLASSIFICATION CONCEPT (ALL DEMAND)
In CHIVE, without regard to security level, different
jobs may be simultaneously in the job queue and the
ready queue, while still another security level may be
held by the job currently in control of the CPU. In other
words, the EDP system will process jobs of different
levels at the same time.
Jobs contending for the CPU are isolated whenever
possible. Such isolation is achieved by means of
safeguards in the Executive Control. The EC employs
the IBM/360 storage protect key, which uniquely specifies
data and programs associated with a given job-task
combination.
SECURITY PROCEDURES
Classification Concept
7.11.3.
-156-
Approved For Release 200gnpf : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/MagiFIRDP78-03952A000100070001-5
A job is classified only because it operates
upon classified data. Therefore, different job-task
classifications exist in the system when and only when
information consumers make simultaneous demands on the
CHIVE data base for different levels of data.
7.11.4. PROBLEM SOLUTION
The combination of the mixed data-classification
concept and the software job/task safeguards in the EC
is the means in principle by which the CHIVE EDP system
solves the fundamental problems posed by the multi-
programming environment. The implementation of this
solution is attained at the cost of added overhead in
the Executive Control and an increased burden on the CPU.
Cost has been held to a minimum consistent with the
no degree of confidence sought. Especially sensitive ad hoc
projects which cannot fit into the security philosophy of
CHIVE in general, must be treated separately. Ad hoc
projects will be reviewed as they arise to determine whether
minor alterations to CHIVE procedures can make them
acceptable.
mei
7.11.5. DECOMPARTMENTALIZED ALL-SOURCE/ALL-SERVICE CONCEPT
Compartmentalization is the technique of achieving
security isolation by physically segregating operational
SECURITY PROCEDURES
Decompartmentalized Concept
7.11.5.
-157-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 20(10i1Th/V8 : CIA-RDP78-03952A000100070001-5
personnel, processor and files, and by timing jobs in
accordance with security level and need-to-know.
Compartmentalization internal to the system is in direct
contradiction to the CHIVE concept of all-source/all-
service.
Decompartmentalization affects all aspects of CHIVE
processing, both within the EDP system and without. From
the standpoint of computer operations, all operators are
qualified to run all jobs. From the standpoint of the
EDP processor, by exercising suitable precautions, jobs
may be intermixed and files may be physically available
on-line irrespective of security. From the standpoint of
information handling operations performed by the informa-
tion analyst, any analyst may service any recipient of
information, placing a signifidant burden for routine
security screening upon the EDP system.
7.11.6. THE ROLE OF "NEED-TO-KNOW" IN CHIVE SECURITY DESIGN
The aim of CHIVE security design has been to achieve
strict security control over outright security breaches
through software as well as hardware safeguards. A
security breach is defined as a violation of security
level, where level is interpreted broadly to include
groups of security classifications such as T/KH/SI/
SECURITY PROCEDURES
Need-to-Know
7.11.6.
-158-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/085K9.RDP78-03952A000100070001-5
Collateral or combinations of these. The security breach
also applied to dissemination caveats, when these are
significant. For example, "CIA Internal Use Only" is
treated as a classification of weight approaching that
of any code word.
When security control has been balanced against
cost, only need-to-know has been affected. Thus, the
accidental writing of one SECRET job output into the
printer assigned to another SECRET job output is treated
as a tolerable risk when safeguards are costly, while
writing a TOP SECRET job output into a .SECRET output is
treated as intolerable.
7.11.7. SECURITY ISOLATION OF PROGRAMS AND DATA IN STORAGE
7.11.7.1. Storage Protect Feature
The storage protect feature is a hardware feature
which significantly affects the handling by the EC of
core blocks of data and. programs. With this feature,
core may not be illegally written.
The EC dynamically assigns matching keys to programs
and data and to blocks of core in accordance with the
task using them. When a program issues an OPEN macro
which calls upon a data file in an I/O device, the EC
SECURITY PROCEDURES
Isolation in Storage
7.11.7.1.
-159-
Approved For Release 2000/05/08 :saltrP78-03952A000100070001-5
Approved For Release 2oo6nme : CIA-RDP78-03952A000100070001-5
will identify the task which it working. It will
then obtain the storage protect key assigned to the
task and assign it to the core block into which the
data will be written. Data read into core from
auxiliary storage by means of a READ macro is
buffered into the block(s) assigned for the use of
this task. The peripheral equipment checks to insure
that the task key matches the key of the affected core
block.
When the program issues a move instruction which
reads and writes data in core, it will use the storage-
protect key given to it by the EC to unlock the data
area. If this key is incorrect, or if the key is
correct but the wrong core block is accessed, an
interrupt will occur. When such an illegal condition
exists, the access will be denied and an appropriate
error message sent to the operator. The error will be
recorded in the system log and the Error Supervisor
will be called.
Resident programs such as the Monitor always will
be assigned to protected core.
SECURITY PROCEDURES
Isolation in Storage
7.11.7.1.
-160-
Approved For Release 2000 I: CIA-RDP78-03952A000100070001-5
sEcRET
Approved For Release 2000/05/CMCM-RDP78-03952A000100070001-5
7.11.7.2. Write-Protect Feature
With this feature core blocks may be read only by
programs having the legal key. Software support parallels
that for the read-only feature.
7.11.7.3. Branch-Protect Feature
A branch protect feature will not be utilized.
Storage-protect and write-protect together obviate the
need for a branch protect, since an illegal branch,
although honored, can only trigger a security breach
when an I/O results, and such a breach is stopped by
the other two features in combination.
7.11.8. SECURITY ISOLATION OF DATA TRANSFERS
Core-to-core transfers of data and programs are
isolated by means of the storage-protect/write-protect
features. The EC finds the key in the Program Status
Word (PSW) at the time the problem program calls for
the move. The key in the PSW is compared by the
circuitry with the key in the core block, which
automatically interrupts when an improper condition
(mismatch) is sensed. The actual transfer is made with
complete parity checking to protect not only the
integrity of the transferred information, but the
routing assignment as well.
-161-
SECURITY PROCEDURES
Isolation of Data Transfers
7.11.8.
Approved For Release 2000/05/08: CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 200?Mripi3 : CIA-RDP78-03952A000100070001-5
Core-to-auxiliary storage transfers and the reverse
are made secure under management of the control programs,
which utilize the protect keys for this purpose. The
data channel uses the storage protect key. The usual
parity checks also are made.
Parity checks are deemed sufficient to secure
peripheral control over routing addresses. However,
routing assignments must be confirmed by the EC in the
manner described below.
When the file is being accessed for the first time
after opening by a specific task, the File Header Record
(FHR)* is interrogated by the EC to insure that the file
on the device is the one sought. This check insures that
the physical file actually accessed on auxilliary storage
is the one called by the problem program and addressed
by the EC.
A sought file may differ from the actual file found
at an auxiliary storage location for these reasons.
(a) In dismountable files, the actual cell or disk pack
mounted on the storage drive may differ from the one
*Note: See Section 5.5.1.5.5 for a discussion of the FHR.
SECURITY PROCEDURES
Isolation of Data Transfers
7.11.8.
-162-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08sEaftrRDP78-03952A000100070001-5
specified. (b) Although unlikely, a file address may
be accidentally altered at issuance time by a machine
error in the storage control unit or in the CPU.
After the first access to a file made by a particular
routine working on a specific task, remaining accesses
are not checked against the FUR. This reduces overhead
without materially increasing risk.
The above checks are made as appropriate for both
file retrieval and maintenance. Writing into files is
checked in order to preclude accidental writing of
sensitive information into less-sensitive files. Although
these checks are very significant security safeguards,
they also are part of the normal checking procedures
built in to the operating system because of other EDP
concerns. Thus, they are available bonus features
whose cost need not be borne by security requirements.
7.11.9. SECURITY ISOLATION OF JOBS AND TASKS
Security isolation of jobs is achieved by assigning
them a security classification. Tasks assume the
classification of the job to which they belong. The
classification of a retrieval job is the highest level of
the sought output. That of an input (maintenance) job is
the level of the most sensitive data being entered.
An explicit statement of job security level is
made by the file analyst in the external EDP environment
SECURITY PROCEDURES
Approved For Release 2000/05/08 : CiA3RDP74- 6110daitibdri1g Tasks
SECRET
Approved For Release 200g6&b : CIA-RDP78-03952A000100070001-5
using the request language. Any job type may involve
any security classification. A specific job of a given
type may have a level differing from that of other jobs
of that type. For example, a specific reference retrieval
job may be ALL-SOURCE, but another reference retrieval
job may be limited to SECRET.
Another form of explicit statement is made in
standing requests, which are held in a table of standing
requests, with output security level an entry. For
standing requests which may take on varying levels,
the user will either specify the level as a parameter in
the request language statement when he calls upon the
standing request, or separate standing requests for a
single job type will be made up for the several security
levels the job type can take on. In the latter case,
each standing request of that type will be referred
to by a different name.
For some job types, there may be only one legal
security level. When parent tasks generate sub-tasks,
the sub-tasks will have a security classification of
their own, since they have an independent life in the
system. This classification can be equal to, less
SECURITY PROCEDURES
Isolation of Jobs and Tasks
7.11.9.
-164-
Approved For Release 20%Mfg98 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/03ECE8A-RDP78-03952A000100070001-5
than or greater than the level of the parent task.
The correct level is determined by any of the criteria
outlined above.
EDP system processing files are not "seen" by the
request writer or referred to explicitly in the request
language, but are required to process a job (see
Section 7.11.10.). These files are called or opened under
control of the Operating System or the problem program.
Hence, opening legality is achieved by correct design of
programs.
7.11.10. PRIVILEGED ACCESS TO LOGICAL FILES
Logical internal files are those files which are
automated within the EDP system. There are two kinds.
Internal information files or data files are files of
data which usually have counterpart files (that is,
Printouts) which they generate for use in the external,
non-EDP environment.
A second kind of internal file is the system
processing file. This kind of file is used by the EDP
system for bookkeeping, logging, system status monitoring
and other purely EDP functions. System processing files
SECURITY PROCEDURES
Access to Logical Files
7.11.10.
-165-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000 ti CIA-RDP78-03952A000100070001-5
may not be directly accessed by the request language.
Access to these files is privileged; it is limited to
CHIVE system programmers.
7.11.11. ACCESS TO AUXILIARY STORAGE UNITS
All programs may have access to any file of data on
any storage unit. After the file (or a block of data
from the file) has been moved into core, it will
reside alone under protection of a storage-protect key.
7.11.12. FILE INTERACTION AND SECURITY ISOLATION
The security level of a data file which is operated
on is not the sole criterion for security isolation.
Two data sets or files under the same storage key may
be vulnerable to security violation even though they are
at an identical level. The combination of source data
security level, data file character (i.e., homogeneous
or heterogeneous), target file level and character, and
task (i.e., file interaction) determine the isolation
criteria.
For example, consider the case in which two TOP
SECRET data files share a storage protect key. They
reside in core. Task I is to screen and convert part of
SECURITY PROCEDURES
File Interaction
7.11.12.
-166-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
owl
oar
Approved For Release 2000/05/06VM-RDP78-03952A000100070001-5
data file I which is heterogeneous into data file IA
which is homogeneous SECRET. Task II is to move data
from file II which is homogeneous into data file IIA
which is homogeneous TOP SECRET. If Task I should
inadvertantly operate on data file II, no screening
will take place, and TOP SECRET data could find
its way into SECRET data file IA.
The File Interaction Security Transition Table
(see Table 7-1) depicts the various combinations of
file interaction. These apply equally to I/O and
core-to-core transfers.
The job classification (as used in Table 7-1) is
the classification for which the information recipient is
cleared. It is the screening criterion, when retrieval
is involved, or it is the input legality criterion
when record maintenance is involved.
Entry type #1 lists the information unit which sets the
level assigned to data units moved out of a file. For
instance, consider a job at a level greater than a
homogeneous source file which moves data out of the file
and into a homogeneous target file which is at a level
beneath that of the job. The level of the moved data
becomes the level of the target file.
SECURITY PROCEDURES
File Interaction
7.11.12.
-167-
Approved For Release 2000/05/08EcRET
: C1A=RDP78-03952A000100070001-5
S
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Table 7-1
FTTN IETTRACTION SECURITY TRANSITION TABLE
Target
One Homogeaeous FileFile
I
One Heterogeneous File
Source
File
Job
Job Class
c1 as
Less Than
Target File
Equal To
Target File
Greater Than
Target File
Less Than
Target File
Equal To
Target File
Greater Than
Target File
' One or
Less
1.
Target File
1.
Source File
More
Than
2.
Nothing
Illegal
Illegal
2.
Extracted Unit
Illegal
Illegal
Homo-
geneous
Source
File(s)
3.
Source File
Target File
3.
Source File
Target File
File(s)
Equal
1.
Target File
1.
Source File/
1.
Source File
1.
Source File
To
2.
Nothing
Target File/Task
Illegal
2.
Extracted Unit
2.
Extracted Unit
Illegal
Source
3.
None
2.
Nothing
3.
None
3.
None
File(s)
3.
None
Greater
1.
Target File
1.
Target File
1.
Target File
1.
Source File
1.
Source File
1.
Source File
Than
2.
Nothing
2.
Nothing
2.
Nothing
2.
Extracted Unit
2.
Extracted
2.
EXtracted Unit
Source
3.
None
3.
None
3.
Source File
3.
Eone
3.
hone
3.
Source File
File
Target File
Target File
Less
1.
Target File
1.
Target File
1.
Target File
1.
Extracted Unit
1.
Extracted Unit
1.
Extracted Unit
One or
Than
2.
Tag Dropped
2.
Tag Dropped
2.
Tag Dropped
2.
Extracted Unit
2.
Extracted Unit
2.
Extracted Unit
More
Source
3.
Soarce Unit
3.
Source Unit
3.
Source Unit
(Tag Unchanged)
(Tag Unchanged)
(Tag Unchanged)
Eetero-
geneous
File(s)
Target File
Task/Target File
Target File
3.
Source Unit
Target File
3.
Source Unit
Task/Target File
3.
Source Unit
Target File
File(s)
Equal
1.
Target File
1.
Target File
1.
Target File
1.
R:tracted Unit
1.
Extracted Unit
1.
Extracted Unit
To
2.
Tag Dropped
2.
Tag Dropped
2.
Tag Dropped
2.
Extracted Unit
2.
Extracted Unit
2.
Extracted Unit
Source
3.
Source Unit
3.
Source Unit
3.
Source Unit
(Tag Unchanged)
(Tag Unchanged)
(Tag Unchanged)
File(s)
Target File
Task/Target File
Target File
3.
None
3.
None
3.
Source Unit
Target File
Greater
1.
Target File
1.
Target File
1.
Target File
1.
Extracted Unit
1.
Task/Extracted
1.
Extracted Unit
Than
2.
Tag Dropped
2.
Tag Dropped
2.
Tag Dropped
2.
Extracted Unit
Unit
2.
Extracted Unit
Source
3.
hone
3.
Lone
3.
Source Tag
(Tag Unchanged)
2.
Ektracted Unit
(Tag Unchanged)
File
Target File
3.
Source Unit
(Tag Unchanged
3.
Source Unit
_
Target File
3.
None
Target File
-168-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
1i I 1 1 ii 1 I
Approved For Release 2000/05/05EGIPPRDP78-03952A000100070001-5
Entry type #2 lists the security level tagging
activity which takes place when the information unit
is moved. It alternatively lists the information unit
which is tagged.
Entry type #3 lists any legality condition which must
be met.
The legality conditions set forth in this table are
followed implicitly by the information analyst in
composing his request. These rules apply to all CHIVE
files. For system-defined files and ad hoc files (working
files which may be destroyed after the job is run), the
analyst specifies and controls legality. For files not
"seen" by the request writer but used by the EDP system
for processing, and for file interactions generated by
sub-tasks, legality is handled internally by the CHIVE
system programs.
In order to simplify the assignment of storage-
protect keys, all data sets called into core by a single
task will be isolated under a unique key.
7.11.13. PRIVILEGED ACCESS TO OUTPUT DEVICES
Output devices are all-source. Protection is gained
by means of physical isolation of the operations area in
SECURITY PROCEDURES
Access to Output Devices
7.11.13
-169-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 26110.105/08 : CIA-RDP78-03952A000100070001-5
which the device is located. Access to areas in which
output devices are located is limited to qualified
(all-source) personnel.
7.11.14. EDP SOFTWARE SECURITY ISOLATION
7.11.14.1. Tamper-proof Operating System
A number of software security features are
implemented by means of the Executive Control.
insure the correct function of these features,
To
it is
necessary to isolate the EC from the action of problem
programs and programmers. This is accomplished by
running the computer in the problem state when problem
programs are in the system. The EC may be modified by
means of privileged instructions which are honored only
when the system is in the supervisory state.
7.11.14.2. Core Erasure
When blocks of core are released by problem programs,
the EC will erase that core by writing zeros in place of
classified data.
7.11.14.3. Auxiliary Storage Erasure
When files of data on fixed auxiliary storage devices
are no longer needed by the system, they will be erased by
-170-
SECURITY PROCEDURES
EDP Software Isolation
7.11.14.3.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08EGRECRDP78-03952A000100070001-5
25X1A
means of a job specially submitted by the computer operator
for this purpose. The job will cause the writing of zeros
on the recording surfaces. Working files not seen by
analysts will be erased under control of the EC. In
this case, the problem program will issue a DESTROY macro
which will be honored by EC.
This form of erasure relates to classified data
stored on on-line storage devices such as disk,
diskpacks and data cell strips. It does not apply to
degaussing of tape.
7.11.14.5. Security Communication with the Computer Operator
In Operating System/360, the EC communicates with the
operator by means of console devices. When a problem
SECURITY PROCEDURES
-171- EDP Software Isolation
7.11.14.5.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release AFC10118i/08 : CIA-RDP78-03952A000100070001-5
program requires access to a secure auxiliary file,
the EC can request the operator to supply the password,
which is needed by the program to open the file. This
feature will not be employed in CHIVE unless extremely
sensitive situations arise.
7.11.14.6. Dumps
When a core dump is requested as the result of an
error condition, the SI/TKH/TS Collateral classification
will be routinely provided at the top and bottom of the
printout. Only the Error Supervisor or error programmer
may initiate a dump; problem programmers or Information
Analysts may not. The Information Analyst who initiated
the job aborted by the error will receive a message
alerting him to the problem, but will not receive the
problem job output.
Dumps will be treated as all-source material, since
the data they contain may be derived from any source.
Appropriate logging of the dump will take place. This
logging will contain lists of files dumped and other
security relevant information.
7.11.14.7. Debug Macro
There will be periods after the system is operational
during which programs are tested and debugged. Such
-172-
SECURITY PROCEDURES
EDP Software Isolation
7.11.14.7.
Approved For Release 2oge9/p8 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/8BalieTA-RDP78-03952A000100070001-5
programs will run in the multiprogramming environment
together with fully tested operational programs.
These undebugged programs may employ a Debug macro
to provide the programmer with information needed for
evaluation and error correction. A Debug macro may
also be utilized from time to time if problems should
arise in operational programs.
To isolate programs using Debug macros, and to
insure that Debug macros do not result in a breach of
security, it is necessary to treat the debug output
material as an all-source file, and to log the print
event.
The printout will contain markings of the security
level of the output (that is, all-source) and a routing
assignment showing the individual or group responsible
for the debugging.
7.11.14.8. Printout of EDP System Processing Files
Request for printout of an EDP system processing
file is privileged. A printout of the system library,
system tables, the decode dictionary, etc., is restricted
to (and needed only by) a CHIVE system programmer or an
EDP File Analyst.
SECURITY PROCEDURES
EDP Software Isolation
7.11.14.8.
-173-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
25X1A
Approved For Release 200(05/#13 : CIA-RDP78-03952A000100070001-5
7.11.14.9. Information Suppression in the Output Phase
Suppression of retrieved material which research
analysts are not authorized to have can take place in the
search phase or in the output phase. Search logic provides
for a search phase suppression. However, it also is
necessary to extend the suppression into the output phase.
Information may be intentionally suppressed in the
output phase for two reasons: (a) a security restriction
may be discovered upon decoding which could not be
discovered on searching, (b) a heterogeneous file may be
screened to produce a report (in report mode) which is
at a level beneath that of the file.
An occurrence of either case may result in the
suppression of the information record, or part of the
record, depending upon the job. The output routines will
contain the logic to determine the suppression option,
and to implement it.
SECURITY PROCEDURES
EDP Software Isolation
7.11.14.10.
-174--
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/054ECRMA-RDP78-03952A000100070001-5
25X1A
7.11.14.11. Remote Operations
Transmission of data to a remote I/O station (such
as a printer not in the computer operating room) must be
accomplished under safeguards of two classes.
Class 1 - Data transmission lines must be secure.
Class 2 - The I/O device must be protected from use
by unauthorized personnel, and the device must be made
secure against tampering. This security control is achieved
by limiting the personnel having access to devices in the
computer room and the operational areas. Procedurally,
users of consoles will erase displays when finished with
them.
-175-
SECURITY PROCEDURES
EDP Software Isolation
7.11.14.11.
Approved For Release 2000/05/01L$16-RDP78-03952A000100070001-5
Approved For Release 20(giatiq8 : CIA-RDP78-03952A000100070001-5
CHIVE EDP system interfaces with other systems
must be safeguarded against improper automatic handover
of information. For example, if the direct control
feature of automatic interruption is employed to mate
the page reader with the CHIVE central processor, the
EC must inspect the direct control command for legality.
7.11.15 EDP HARDWARE SECURITY
7.11.15.1. Hardware Security Levels
All hardware subsystems will be assigned a "lifetime"
security level'of ALL SOURCE signifying the all-source
jobs or data with which they may be associated.
CHIVE-internal fixed physical storage files used with
hardware also will be all-source. Dismountable files
sent to external agencies will be classified in accordance
with the data they contain.
7.11.15.2. Input/Output Hardware
I/O devices located within the EDP facility are
lifetime cleared for all-source data.
A display user will erase (if possible) when he is
finished using a device.
SECURITY PROCEDURES
EDP Hardware Security
7.11.15.2.
-176-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/MCCEM-RDP78-03952A000100070001-5
Any I/O devices located outside the computer facility
are routed through a patch panel or similar physical
or electronic disconnect device, so that they may be
physically and/or electronically isolated from the EDP
system when not in use. The devices will be manually
enabled at the start of a use period, and disenabled
afterward.
7.11.15.3. Auxiliary Storage Devices
These are of two kinds: permanent and dismountable.
Permanent storage devices are devices such as disks
which are kept on-line most or all of the time. When
the recording surface is removed, a modification to the
drive is required, or the entire drive itself must be
removed. Dismountable devices are designed to facilitate
removal of the recording surface, as in the case of a
diskpack, tape or cell on a strip file.
The lifetime level of the drive and the recording
surface may differ. They may differ when the dismountable
surface is routed to an agency other than CHIVE. The
lifetime security state of the recording surface of a
permanent device is the same as that of the drive.
SECURITY PROCEDURES
EDP hardware Security
7.11.15.3.
-177-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 200?tiZit0f3 : CIA-RDP78-03952A000100070001-5
Dismountable devices are color coded in accordance
with the security level of the data currently recorded
thereon. They also bear a lifetime code, to indicate
whether or not they have been released outside of CHIVE.
Such a releasable surface also bears a code to
indicate the most sensitive data it has borne since
the last complete erasure (in compliance with degaussing
and downgrading regulations). It is treated in handling
in accordance with the most sensitive of the two following
criteria: either the data it now bears or the data it has
borne since complete erasure.
They also are coded in a color system compatible
with that used by the outside agency. They may be
mounted on-line while CHIVE all-source files are on-line.
Data from CHIVE may be written directly on these devices.
If data recorded on external files is to be brought
into CHIVE, it will be recorded in a CHIVE-compatible
file format. That is, File Header Records must be
available on the recording surface.
7.11.16. PROCEDURAL SECURITY
7.11.16.1. Definition
Procedural security is defined to encompass security
procedures relating to computer control room operations.
SECURITY PROCEDURES
Procedural Security
7.11.16.1.
-178-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05fttRAIA-RDP78-03952A000100070001-5
7.11.16.2. Decompartmentalized Operating Cony.p...2L
All CHIVE computer operating personnel are cleared to
run all FDP jobs. The output of every printer or other
I/O device in the computer area may be inspected by any
operator as part of his assigned duties. Al]., equipment in
the control room is open and unscreened. Physical
security is considered an adequate barrier to the
prevention of unauthorized release of information in the
computer facility.
7.11.16.3. Operator Responsibility
The Executive Control is designed to free the computer
operators from undue routine burdens, and at the same
time to make them an integral part of the system. For
example, in Operating System/360 an operator could be
asked by EC to supply a password needed to open a secure
file.
In addition to their usual responsibilities,
operators are expected to monitor the proper operation
of the system with respect to security functions.
Examining printer output, for example, is a desirable check
which operators can make to maintain watch over the system.
Insuring that output is routed to the proper recipient is
another.
-179-
SECURITY PROCEDURES
Procedural Security
7.11.16.3.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
25X1A
Approved For Release 20010M8 : CIA-RDP78-03952A000100070001-5
7.11.16.4. 211LERL Routing
All CHIVE output destined for research analysts
(that is, all output which leaves CHIVE per se or is
exposed to non-CHIVE personnel) must be screened by an
information analyst. No request can be written by
other than an authorized member of the CHIVE staff.
Some authorized offices within the government will
issue standing requirements for information. These
will be expressed in standing requests and might provide
a routing destination which is printed out together with
the answer. When authorized by the Controlling Information
Analyst, the operator will use this routing to transmit
the copy to the originating office.
Output destined for information analysts will also
contain routine information, including the name of the
Information Analyst and the name of the research analyst
(if any). Responsibility for release rests with the
Information Analyst.
SECURITY PROCEDURES
Procedural Security
7.11.16.5.
-180-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/050tRCIA-RDP78-03952A000100070001-5
25X1A
7.11.16.6. Environmental Isolation
Environmental ioslation refers to security aspects
of CHIVE interfaces with other systems, which comprise
the environment.
When interfacing systems are steady customers of
CHIVE, or when they frequently contribute to the CHIVE
data store, a file circulation system will be set up.
Storage devices which are sent to customers containing
CHIVE data will be returned for reuse.
-181-
SECURITY PROCEDURES
Procedural Security
7.11.16.6.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 20VALKO8 : CIA-RDP78-03952A000100070001-5
7.11.16.7. Printouts
The machine room will retain no copies of printouts.
Codeword printouts will not be logged by the machine room.
Printouts are logged manually by analysts only if
released directly to CHIVE customers and only for TKH and
TS Collateral material. Offices originating requests will
be responsible for logging the output received from CHIVE.
Logging will be performed by the cognizant security
office(s).
Information analysts may destroy any unlogged
printout without logging the destruction. Printouts are
treated as classified waste within CHIVE boundaries.
7.11.16.8. CHIVE Vault
A vault in the computer area is used to safeguard
CHIVE machine files. The file library is maintained by
a CHIVE librarian, who logs and controls the use of
files.
7.11.16.9. CHIVE Courier Service
Couriers are used to transport printouts and
documents between the computer room and the service desk(s),
which are not behind the same barrier.
SECURITY PROCEDURES
Procedural Security
7.11.16.9.
-182-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/CalCIM-RDP78-03952A000100070001-5
25X1A
7.11.17. COMPUTER FACILITY SECURITY; PREMISES SECURITY
Over-the-counter service will be provided to
customers in a secure area separated from the computer
area. Desks providing service to one type of customer
(that is, a customer cleared for one combination of
codeword/collateral) will be isolated from those
providing service to another type.
7.11.18. INFORMATION HANDLING FUNCTIONS AND OPERATIONS
7.11 18.1. Reliance on Personnel
Maintaining security control in CHIVE is a joint
man/machine undertaking. Screening of output by
information analysts is supported by EDP techniques
intended to reduce the manual burden.
-183-
SECURITY PROCEDURES
Information Handling
7.11.18.1.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release gatlfa5/08 : CIA-RDP78-03952A000100070001-5
7.11.18.2. Information Analyst
Service will be provided to customers who fall
into the following classes.
- TKH/SI/Collateral all levels
- SI/Collateral all levels
- Collateral all levels
- Collateral SECRET and below
Dissemination caveats will be observed within each group.
The Information Analyst is expected to determine
the customer's classification and need-to-know. He is
expected to reflect these items in EDP requests,
providing indicators in the request language which the
EDP system may use to determine screening requirements.
He is expected to exercise options in stipulating the
desired output (records, documents, etc.) in accordance
with security requirements.
The Information Analyst will also security control the
dissemination of microimages, printouts, hard copies of
documents, etc. He will often provide information retrieval
directly to the customer out of prepared printouts, imposing
screening restrictions at the time of service. He is
required to keep records of the release of secure informa-
tion and/or documents.
-184-
SECURITY PROCEDURES
Information Handling
7.11.18.2.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/98CRETA-RDP78-03952A000100070001-5
7.11.18.3. File Analyst
The File Analyst defines the content and structure
of CHIVE EDP files. He also defines content and
structure of the security aspects of files, including
their homogeneity or heterogeniety, levels or ranges
of data, caveats, etc. These are used by the EDP
system programs to interpret requests, determine
their legality, and to perform checking of various kinds.
No file may be defined without a security statement.
The File Analyst verifies the file definitions by
examining a machine-produced copy of the File Format
Table. This insures that the EDP system has properly
accepted the definition.
7.11.18.4. Document Control and Security Officer
A CHIVE officer will be appointed to assist in
procedural security. He will exercise control over
procedures regarding printouts and documents which are
input or output from the EDP system.
In performing CHIVE security functions, the
security officer will not represent the cognizant
security offices. Instead, he will perform liaison
with these offices.
SECURITY PROCEDURES
Information Handling
7.11.18.4.
-185-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/af%R-RDP78-03952A000100070001-5
Chapter 7.12.
BACKUP PROCEDURES
7.12.1. BACKUP CONCEPTS
File duplication is the primary means by which
CHIVE data or programs are preserved against complete
loss. In the EDP system, both data files and system
processing files are preserved.
No backup system is foolproof. A backup system
seeks a high degree of confidence in protection against
complete loss due to handling damage, loss, damage
due to program bugs, machine damage, disaster, etc.
Confidence may be achieved by a combination of
precautionary methodology and recovery techniques.
Manual and automatic means are available to reduce the
likelihood of total loss. The tape file protect ring
is an instance of manual prevention; EDP file opening
legality checking is an instance of automatic prevention.
Recovery consists of maintenance or resumption of
normal system operations after information loss due? to an
extraordinary incident by use of reserve copies of files,
by reconstruction of files, or by transformation of data
BACKUP PROCEDURES
Concepts
7.12.1.
-187-
Approved For Release 2000/05/08EWDP78-03952A000100070001-5
Approved For Release 200205818 : CIA-RDP78-03952A000100070001-5
stored in one form or medium into another form or
medium more suitable to the attainment of system goals.
Resources committed for backup are usually
limited by the expectation of infrequent system breakdown
due to failure. The backup design is a compromise between
reduced performance during down periods and the relative
frequency of reduced performance. In CHIVE a fall-back
concept has been employed. At a time of loss, operations
are not crippled although capabilities are reduced. The
severe cost burden of striving for fu/1 performance
capabilities during temporary periods of failure thus
is not borne.
In the CHIVE fall-back approach, the chosen temporary
cost may consist of some system feature given up
entirely, or some degradation in performance, or a combina-
tion of these. For example, it may be decided to suspend
service to an unimportant customer class, or to provide
this service with a longer response time. Instead of using
an EDP file to perform a search, a printout may be used.
7.12.2. BACKUP PLAN
There are two versions of a backup capability for
CHIVE. Version I is implemented in the initial system.
It represents a simplified low-cost solution which will
BACKUP PROCEDURES
Plan
7.12.2.
-188-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0gctEiA-RDP78-03952A000100070001-5
meet the limited preliminary system needs of the first
period. Version II represents the full backup capability
which will be implemented gradually. Both versions are an
integrated complex of human procedures, EDP software
features, and hardware capabilities.
There are two fundamental procedures in the file
backup for the Version I plan. First, the entire system
is dumped once per month, with the exception of files
which have had no activity during the period. Secondly,
all transactions for the current month are saved in
their original form against a possible loss of the current
monthly file.
If lost, the previous monthly file is used in
conjunction with the saved transactions to build back
another current file. Multiple copies of the dump are
available in hard copy form on printouts in case the
magnetic form of backup is lost. These copies are
dispersed against possible disaster.
Standard office policy is established to determine
the handling procedures for duplicate copies. These
procedures are developed jointly by OCS and OCR. Some
special procedures may be developed for certain classes
of file. For example, slowly changing authority files
BACKUP PROCEDURES
Plan
7.12.2.
-189-
Approved For Release 2000/05/08siEIINRDP78-03952A000100070001-5
Approved For Release 2009M60 : CIA-RDP78-03952A000100070001-5
may be backed up and built back differently than
rapidly changing personality summary files. When an
information analyst creates a new permanent (not working)
file, he is responsible for declaring that the file is a
system file of a said class. The procedures for backup
are then automatically instituted throughout the CHIVE
system for that file.
Hardware backup is achieved in Version I primarily
by utilization of the existing OCS computer facility.
Storage units (drives) will be shared by the two organiza-
tions. Multiple storage units will be available in case
of the failure of a single unit. In Period Two (that is,
roughly in the second year of operation) CHIVE files will
grow to a size which requires a special CHIVE strip file.
This file will be backed up by another drive used by OCS.
Should a failure occur in either file, the data cells on
both can share the operating drive.
One of the most important CHIVE backup techniques is
based on the page reader (PR). From a cost-effectiveness
point of view, the combination of printout dumps of files
and the PR is one of the most attractive forms of backup,
due to the low cost of paper as a recording medium and the
ease with which multiple hard copies may be made. Hard
BACKUP PROCEDURES
Plan
7.12.2.
-190-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/MGRVA-RDP78-03952A000100070001-5
copy recording will be used to supplement the usual
magnetic tape recording for backup. The first generation
copy will be stored on tape; succeeding generations will
be stored on printouts.
The Version II plan is a developmental one. As the
size of CHIVE files increases, more sophisticated backup
techniques will be needed. For example, duplicate copies
of incoming index records may be built on-line on
different strips of data cell drives.
The phase-over from the initial backup plan to a
more sophisticated plan will be gradual. The principal
criteria for instituting changes in the initial plan will
be (a) reduction of backup costs, (b) increasing operational
capability during recovery, (c) increasing confidence
that build back capability exists to several generations
of file for all forms of loss, (d) responding to
increasing 7erformance requirements under increasing
loads, (e) achieving additional capabilities.
BACKUP PRO=URES
Plan
7.12.2.
-191-
Approved For Release 2000/05/0%ia*RDP78-03952A000100070001-5
Approved For Release 2000/05/%dgiVRDP78-03952A000100070001-5
Appendix 7.A.
USER LANGUAGE MANUAL
Approved For Release 2000/05/08 iReDP78-03952A000100070001-5
Approved For Release 2cO/R4O8 : CIA-RDP78-03952A000100070001-5
Appendix 7.A.
USER LANGUAGE MANUAL
7.A.1. PURPOSE OF MANUAL
Appendix 7.A. contains the preliminary specifications
of the CHIVE language for querying and maintaining the
CHIVE data base.
Briefly, Appendix 7.A. deals with:
- A description of the logical data structures
(CHIVE files, records, phrases, and terms)
which can be addressed and manipulated in the
language.
- A brief description of the general capabilities
of the language from the standpoint of the
user (analyst). That is, a general description
is given of what the language can do to assist
the analyst in the solution of information
processing problems.
- A specification of the rules (grammar and syntax)
which will guide the analyst in correctly
formulating his requests. The specification of
rules constitutes a description of how expressions
in the language can be formed to perform various
EDP tasks.
The rules for specifying retrieval transactions are
presented first. These are followed by the rules for file
maintenance transactions.
To demonstrate the capabilities of the language and
to illustrate the use of the rules, examples are stated
USER LANGUAGE MANUAL
Purpose of Manual
7.A.1.
-192-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/0MOORECIA-RDP78-03952A000100070001-5
in ordinary English and are followed by a formulation
of the problem in the CHIVE language.
7.A.2. CHIVE DATA ELEMENTS AND THEIR LOGICAL STRUCTURE*
The CHIVE language is designed for use with a
generalized logical data structure which is user oriented:
all CHIVE EDP records are similar in their logical
structure to the structure of the document index record
410 as it is developed by the analyst in the document
indexing process. Moreover, the user can think in
dr
terms of a hierarchic structure of terms, phrases, and
records in each file. The structure is identical for a
file of document index records and for information files
mi
of any functional type (e.g., vocabulary lists, summary
files, and unsynthesized files). As far as the user is
concerned, the records in all of these files will be
serially searched and scanned.
In order to exhibit this logical data structure,
several logical levels of data can be distinguished and
IMO
aul0
identified as follows.
*Logical data structures are distinguished from the
physical data structures of the system; the latter term
refers to the way data is organized and represented
internally or "in the machine." The former term denotes
the conceptual framework to be used in the analysis and
specification of information problems in the CHIVE language.
USER LANGUAGE MANUAL
CHIVE Data Elements
7.A.2.
-193-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 206011M8 : CIA-RDP78-03952A000100070001-5
7.A.2.1. CHIVE Files
A CHIVE file isa named collection of records which
can be processed as input to the CHIVE EDP System. This
includes intermediate files that are dynamically developed
by the user (information analyst) at execution time.
These are temporary collections of records which are
developed for subsequent processing in the same retrieval
transaction.
7.A.2.2. CHIVE Records
Records can be described, first, in terms of two
levels within a record: header and content levels;
and secondly, in terms of phrase groups and phrases.
7.A.2.2.1. Header and Content Levels
The user can access and manipulate data at each of
these levels within the record. The header phrase
must include one or more record control terms which
uniquely identify each record. For example, an information
file might be controlled on PNO* header terms. In this
case, each record in this information file would contain
information on a unique personality. As another example,
*See Volume V for tag definitions.
USER LANGUAGE MANUAL
CHIVE Data Elements
7.A.2.2.1.
-194-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0/tcalA-RDP78-03952A000100070001-5
the record control for document index records is a combination
of the accession number term and the item number term.
7.A.2.2.2. Phrase Groups and Phrases
The content level of each record is made up of
content phrases which belong to phrase groups. Each content
phrase logically links together a set of terms into
some logically meaningful combinations. Figure 7.A-1
shows a sample CHIVE record with two phrase groups
within the content level. A phrase group is made up
of one or more content phrases, all of which may
contain terms with a set of tags that is unique to
that group. Each tag of a particular phrase group
may occur in any phrase of that group. However, it
cannot occur in a phrase of any other phrase group
in the records of a particular file.
VJ_gure 7.A-2 shows in more detail the header and
content levels of the above sample CHIVE record. The
content level contains phrases belonging to the two
phrase groups. Phrase group #1 contains terms with tags
CDY, CNC, and CSS. It contains three content phrases
defined in this way. Phrase group #2 contains terms
USER LANGUAGE MANUAL
CHIVE Data Elements
7.A.2.2.2.
-195-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Header Level
0 I
Ill I-a
7, as
-4 Content Level
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Figure 7.A-1
Logical Structure of CHIVE Records.
HEMER PHRASE
Term
Tag-value
CONTENT PHRASE GROUP # 1
Term
Tag-value
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
1 ii 1 1 1 1 1 1 1 1 1
Phrase
Group
#1
Phrase
Group
#2
I Li t t t
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Phrase 1
Phrase 2
Phrase 3
Phrase 1
Phrase 2
Phrase 3
Phrase 4
Figure 7.A-2
SAMPLE CHIVE RECORD
HEADER
PHRASE TERMS
CDY
Value
CNC
Value
CSS
Value
CNC '
I Value
CDY
Value
CNC
Value
CSS
Value
I
CDY
Value
CNC
Value
CSS
Value
OLC
1
ONP
ONR
Value
1 Value
Value
OLC
ONP
1
ONR
Value
Value
1 Value
OLC
ONP
ONR
Value
Value
Value
OLC
Value
ONR
Value
Content Level of CHIVE Record with Two Phrase Groups
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000510aMt: CIA-RDP78-03952A000100070001-5
with tags OLC, ONP, and ONR. It contains four content
phrases in this way. Notice that the first phrase of
phrase group #1 contains two CNC tags. Thus, although a
tag cannot occur in more than one phrase group, it may
occur several times within an individual phrase of its
particular phrase group.
Notice also that in the last phrase of phrase
group #2, no ONE' tag occurs. Thus, although a phrase
group is defined by a particular unique tag set, any
tag of that set may or may not occur among the terms
of a specific phrase.
It is for the above reasons that the content level
of a document index record is said always to contain
content phrases of one and only one phrase group. All
phrases of a document index record belong to this one
phrase group. This phrase group is defined by the set
of all the tags in the document index language.
Figure 7.A-3 shows in detail the logical structure
of a document index record*. Figure 7.A-4 shows a detail
of the logical record structure of an information record.
*Field positions for header information might not, of
course, correspond to those given in the examples.
USER LANGUAGE MANUAL
CHIVE Data Elements
7.A.2.2.2.
-198-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
mai
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Figure 7.A-3
SCHEMATIC AND TREE REPRESENTATION
OF
DOCUMENT INDEX RECORD
Header
Phrase
Phrase
#1
Content
Phrase Phrase
Group #2
Phrase
#3
Content Phrase Group
Phrases
Terms
Record Control
Term
.."-
Tag ' Classifica -
Value 1
1 1
tion
LAC
1393
LAG
i 3206
LAR
7971
PNO
LAF
67021
PLC I
1013
PDT'
062832 062832 1
PDD I
1201049
PDT
1 041264
11111
PDL 11111
090163
PA
Header Phrase
-199-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
25X6
25X6
25X6
Approved ForRelease2000/05/0VOSIN-MP78-03952A000100070001-5
SCHEMATIC AND TREE REPRESENTATIONS
OF
INFORMATION RECORD
Header
Phrase
Phrase {Phrase #1
Group 1
(Phrase #1
Phrase #2
Phrase
Group Phrase #3
2
Phrase #4
Phrase #5
Phrase Phrase #1
Group
3 IPhrase #2
Phrase
Groups
Phrases
Terms
Record
PNO 1
Classifi-
PLP 1
PDP
Control
cation
13449
i 0802-
PEO 273
PES
29
PED 7 PDE
PFN
3113
PFM
3
PFO
27
PFH
PFN
7612
PFM
3
PFO
22
PFH
PFN
0174
PFM
3
PFO
22
PFH
PFN
6920
PFM J
3
PFO
22
PFH
PFN
5391
PFM
3
PFO
27
PFH
PJP
27
PHT
39
PHR
49
PHN
PJP
39 PITT 94 PHR
77 PHN
Header Phrase
Approved ForRelease2000/05/08 : CIA-RDP78-03952A000100070001-5
I-111111111111a a
25X6
MI
ant
-rail
Awl
Approved For Release 2000/05AW@IA-RDP78-03952A000100070001-5
Note that a phrase group might contain only one
phrase member as in the example of Figure 7.A-4. Here,
Phrase Group #1 has as its only member Phrase #1.
7.A.2.2.3. Terms
The lowest logical level of CHIVE data elements is
the term. It consists of a tag and a value. The
following parts of a term can be examined by the system
and can thus be accessed in the language:
- The whole term including tag and value
- The tag alone
- The value alone
7.A.3. GENERAL ORGANIZATION OF CHIVE LANGUAGE
7.A.3.1. Broad Functional Capabilities
The CHIVE user language has broad functional
capabillties in the two major areas of file maintenance
and information retrieval. Maintenance capabilities
consist of the unconditional and conditional modification
of the data (tags and values) in the records, phrases
or individual terms of CHIVE system data files. Retrieval
USER LANGUAGE MANUAL
General Organization
7.A.3.1.
-201-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000=98T: CIA-RDP78-03952A000100070001-5
capabilities include:
- Accessing files and searching records to retrieve
phrases and terms
- Modifying the retrieved data elements
- Sorting and reordering the records in a file
- Writing new system data files
- Printing output reports containing the retrieved
data
The CHIVE language has been designed for use by the
information analyst. There are several other important
EDP functions performed by the file analyst which are
not handled by this language. These functions are
generally handled by IBM/360 assembler or compiler
languages as described in Volume VII.
7.A.3.2. Job Specifications
The IBM Operating System/360 processes individual
input entities called "jobs." Each CHIVE job can consist
of one or more maintenance transactions, or one or more
retrieval transactions. Following is an outline of the
specifications which must be provided with each job:
Maintenance Job Specifications
- Requestor
- Editor
- Classification
- Type of job (retrieval or maintenance)
USER LANGUAGE MANUAL
-202- General Organization
7.A.3.2.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05MRECIA-RDP78-03952A000100070001-5
- File Name
- Date
- Priority
Retrieval Job Specifications
- Requestor
- Customer (name and component)
- Classification (overall)
- Type of job (retrieval or maintenance)
- Indication of whether or not File Format Table control
is to be exercised
- Date
- Priority
7.A.3.3. Input Coding Sheet
Figure 7.A-5 shows the coding sheet on which the
user defines maintenance or retrieval transactions.
The coding sheet contains test lines and command lines.
In test lines the user specifies test conditions whose
true or false results identify those records, phrases,
or terms which are to be operated on by the commands of
the CHIVE language. In command lines, the user specifies
the various types of maintenance and retrieval commands
in the CHIVE language. Many of these commands will
reference test lines to determine the record, phrases or
terms to be operated on.
-203-
USER LANGUAGE MANUAL
General Organization
7.A.3.3.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Figure 7.A-5
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
TEST AND COMMAND LINES ON INPUT CODING SHEET
FUNCTION
1. Test Condition
Specification
2. Information
Process Direction
USER FORMAT
1. Test Lines
2. Command Lines
N
TEST
TAG
OP
VALUES
A
B
C
D
E
F
G
H
I
J
K
L
COMMAND
LIMIT
CRITERIA
R P T
FILES
IN OUT
DATA
Approved ForRelease2000/05/08 : CIA-RDP78-03952A000100070001-5
I 1 1 1 111IIII
Approved For Release 2000/05/MICTA-RDP78-03952A000100070001-5
7.A.3.4. Test Conditions
Each test line specifies one logical test condition
whose name is given in the TEST column as one of the
letters:
Al BI Cf40.0,
Y. Each test condition has two
possible results--true or false. There are two types
of test conditions--simple and complex. A simple test
condition is the basic logical expression written on one
line. Complex test conditions reference other test
conditions (either simple or complex), so that the true
or false result of a complex test condition is a function
of the true or false result of its constituent test
conditions.
7.A.3.4.1. References to Test Conditions in the CRITERIA
Field of Commands
The names, (A, B, Y), of the test conditions are
referenced in the three CRITERIA fields of command lines:
? = record level search criteria
? = phrase level search criteria
"T" = term level search criteria
CRITERIA test conditions which are true determine
the operands of the CHIVE commands for the following purposes
(see Table 7.A-1 and Section 7.A.3.5.):
- Determine those records (R), phrases (R and P) or
terms (R, P and T) which are to be retrieved from
a named input file (COPY, EXTRACT and EXPLODE
commands)
USER LANGUAGE MANUAL
-205- General Organization
7.A.3.4.1.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2006100 : CIA-RDP78-03952A000100070001-5
- Determine those records (R), phrases (R and P) or
terms (R, P and T) which are to be modified by the
file maintenance or retrieval hit processing
commands (APPEND, DELETE and REPLACE commands)
- Determine those records (R),
terms (R, P and T) which are
computational and definition
ACCUM and DEFINE commands)
phrases (R and P) or
the operands of
commands (TALLY,
The special letter "Z" in an R, P or T Search
Criteria field indicates that all the records, phrases
and terms respectively which are examined are qualified
as operands for the command. "Z" is equivalent logically
to specifying the name
always true.
The user can conceptually think that the true or
false result of the CRITERIA test condition is evaluated
for _very record of the input file specified in the command.
The internal EDP processing may be such that the
specified CRITERIA test conditions are not evaluated
for every record, phrase or term. Indeed the use of
devices such as directory files is aimed at avoiding the
evaluation of test conditions on all records in a system
data file. Nevertheless it is most convenient in using
the CHIVE language to imagine that the R, P and T test
conditions are exhaustively evaluated as described above.
of a
test condition which is
7.A.3.4.2. Simple Test Conditions
The most common simple test condition is of the
USER LANGUAGE MANUAL
-206- General Organization
7.A.3.4.2.
Approved For Release 2CORNO8 : CIA-RDP78-03952A000100070001-5
rod
Approved For Release 2000/05fieggIA-RDP78-03952A000100070001-5
following kind:
TEST TAG OP VALUES
A TAGNAME 1 TERM VALUE
Expressed in ordinary English, simple test condition A
is true if the record or phrase currently being examined
has a term with the tag "TAGNAME" and has at lease one
value equal to "TERMVALUE." If no term can be found in
the record or phrase with the specified tag and value,
then test condition A is false with respect to that record
or phrase.
Other types of simple test conditions can be
specified by an appropriate entry in the OP field as
explained in Section 7.A.4.2. These include other
relational tests besides the equality test described
above For example, does the record or phrase currently
being examined have a term with the tag "TAGNAME" and
with a value that is less than or greater than the
value "TERMVALUE" specified in the test condition. A
span test condition would have two values (TERMVALUE 1
and TERMVALUE 2) specified. It would be true if the
record or phrase currently being examined has a term
with the tag "TAGNAME," and has a value which is
USER LANGUAGE MANUAL
General Organization
7.A.3.4.2.
-207
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2006105MIT : CIA-RDP78-03952A000100070001-5
9reater than the first test condition value TERMVALUE 1
but less than the second value TERMVALUE 2. A scan
test condition would specify a combination of required
characters and "don't care" characters in the VALUE
field. A scan test would be true if the record or phrase
currently being examined has a term with the tag "TAGNAME"
and has a value which contains the specified string of
required characters and don't care characters.
7.A.3.4.3. Complex Test Conditions
The true or false result of a complex test condition is
a logical function of the true or false results of
specified simple test conditions. Consider for example
the following complex test condition:
TEST TAG OP VALUES
3 G, F, Q, R
Expressed in ordinary English, complex test
condition H is true if at least 3 of the four test
conditions (G, F, Q, and R) are true anywhere in the
record currently being examined. These four test conditions
may be either simple test conditions or other complex test
conditions. In general, the OP field of a complex test
USER LANGUAGE MANUAL
General Organization
7.A.3.4.3.
-208-
Approved For Release 29g9jlii/08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/OWNREVA-RDP78-03952A000100070001-5
condition contains a number M, while the VALUES field
specifies the names of N test conditions. The complex
test condition is true if M or more of the specified
N test conditions are true. Section 7.A.4.3 describes
complex test conditions in more detail.
7.A.3.4.4. Phrase Coordination of Test Conditions
The special asterisk character "*" can be inserted
in the VALUES field of a complex test condition as shown
below to indicate that the M of N test conditions must be
true in the same phrase for each record or phrase being
examined.
TEST TAG OP VALUES
3 * G, F, Q, R
Expressed in ordinary English, complex test condition
K is true only if at least 3 of the four test conditions
(G, F, Q and R) are true for a given phrase in the
record currently being examined. Test condition K
must therefore logically be evaluated for each phrase in
the record. Without the "*" test condition K would be
evaluated only once across the entire record, with no
USER LANGUAGE MANUAL
General Organization
7.A.3.4.4.
-209-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 200SEME18 : CIA-RDP78-03952A000100070001-5
regard for phrase linkage. Test condition K is
obviously more stringent than H, since H is true if
three or more of the test conditions G, F, Q, and R
are true anywhere in the record, and not necessarily
in a single phrase.
7.A.3.5. Command Lines
Command lines contain the following eight fields:
COMMAND LIMIT CRITERIA FILES DATA
R P T IN OUT
Table 7.A-1 summarizes the manner in which these
fields must be coded for each command. There are three
possibilities:
- "yes" indicates that the appropriate operand(s)
must be entered in the field
- "no" indicates that no entry is required. Fields
with "no" will simply be ignored
- "opt" indicates that the field is optional. A
blank field is therefore as significant as a
non-blank entry.
7.A.3.5.1. Summary Description of Command Lines
Commands are generally performed on a named Input
File. The data elements operated on are specified by
USER LANGUAGE MANUAL
General Organization
7.A.3.5.1.
-210-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05EMECIA-RD1278-03952A000100070001-5
Table 7.A-1
Summary of Rules for COMMAND Lines
COMMAND LIMIT CRITERIA
FILES DATA
IN OUT
COPY
Opt yes no no yes
Opt yes yes no yes
yes yes yes yes
yes no
yes no
yes no
EXPLODE Opt yes yes no yes yes no
no
Opt yes yes yes 7CS res
EXTRACT Opt yes no no yes no no
Opt yes yes no yes no no
Opt yes yes yes yes no no
APPEND Opt yes no no no no yes
Opt yes yes no no no yes
REPLACE Opt yes yes no no no yes
Opt yes yes yes no no yes
DELETE Opt yes no no no no
yes yes no no no no
yes yes yes no no no
TALLY r opt yes no no yes no no
Opt yes yes no yes no no
Opt yes yes yes yes no no
Terms or
$TALLY n
$COMPUTE n
$ACCUM n
$DEF1NE n
DEFINE n no yes yes no yes no no
yes yes yes yes no no
PRINT no no no no yes yes yes Sort key;
report
parameters
ACCUM n Opt yes yes yes yes yes yes $TALLY n
$ACCUMb
$COMPUTE n
MERGER Opt no no no yes yes no
MERGE P Opt no no no no no yes
WRITE P Opt no no no no yes no
WRITER OPT no no no no yes no
COMPUTE n Opt no no no yes no yes
Approved For Release 2000/05/08441A-RDP78-03952A000100070001-5
SECRE
Approved For Release 2000/650NET CIA-RDP78-03952A000100070001-5
entering test condition names in the following Search
Criteria fields:
- Records (R field only)
- Phrases (R and P fields)
- Terms (R, P and T fields)
A number may be placed in the Limit field which
specifies the maximum number of times that the command
is performed on the particular data elements of the
input file. The results of the command will generally
be written on a named Ouput File. Some commands require
the specification of additional operands or parameters
in the Data field.
7.A.3.6. FFT Control Concept
A retrieval job can be conducted in two basic modes:
- Under File Format Table control
- Without File Format Table control
FFT control is the normal mode. Under FFT control
the Test Conditions, and the Value Fields of Commands can
only reference tags which have been defined in the File
Format Table of the named Input File. In addition the
APPEND, DELETE and REPLACE commands are subject to
certain restrictions under FFT control as described in
Section 7.A.4.7.
USER LANGUAGE MANUAL
General Organization
-212- 7.A.3.6.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05386R1311A-RDP78-03952A000100070001-5
When a retrieval job is not under FFT control,
any arbitrary tags can be referenced in the Test
Conditions or the Data Fields of Commands. Most
of the well defined restraints imposed by the FFT are
relaxed.
7.A.4. LANGUAGE SPECIFICATIONS
7.A.4.1. Summary of Retrieval Commands
Many retrieval transactions require only the two
basic retrieval commands--COPY and PRINT. The COPY
command copies those records, phrases or terms which
meet the search criteria. These data elements are copied
from a named Input File. The PRINT command prints a
named Input File (generally the COPY Output File) according
to a named output report format. The EXPLODE command is
similar to the COPY Command except that each phrase or
term in the named Input File which meets the specified
Search Criteria is copied as an individual record on the
named Output File.
The COPY command copies retrieved data elements
directly into a named output file. When it is necessary
to edit or modify hit data elements, the EXTRACT and
WRITE combination must be used. Included between these
two commands are four possible commands--APPEND, DELETE,
USER LANGUAGE MANUAL
-213- Language Specifications
7.A.4.1.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2009M013 : CIA-RDP78-03952A000100070001-5
REPLACE and MERGEP which modify the retrieved data.
The EXTRACT command extracts records, phrases and
terms which meet the search criteria into a temporary
work area. This extracting occurs one record at a time.
A succeeding WRITE command will write out the modified
contents of the work area into a named Output File. The
APPEND command appends data which appears in the Data
Field to the phrase(s) or term(s) in the work area which
meet the Search Criteria. The DELETE command deletes
data elements which meet the specified Search Criteria
from the record in the temporary work area. The REPLACE
command replaces the data elements in the work area
record which meet the Search Criteria with data specified
in the Value Field.
There are three computational commands--TALLY,
ACCUM and COMPUTE--which perform computational operations
on CHIVE data elements. TALLY counts the number of
records, phrases or terms in a named Input File which
meet the specified Search Criteria. ACCUM accumulates
the sum of the numeric valued terms in a named Input File
which meet the specified Search Criteria. COMPUTE
performs arithmetic operations (add, subtract, multiply
and divide) on CHIVE data elements and on user-defined
values.
-214-
USER LANGUAGE MANUAL
Language Specifications
7.A.4.1.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/UENETCIA-RDP78-03952A000100070001-5
The MERGER command causes all records in a named
input file, which contain a common term(s) to be
physically co-located in a named output file. The
MERGEP command must be preceded by an EXTRACT command
and hence it operates on the single extracted record
in the temporary work area. All phrases within the record
with a common value for a specified merge tag are merged
into a single new phrase.
7.A.4.2. Simple Test Lines
A simple test line is made up of the following
entries:
- TEST
-TAG
- OP
- VALUE(S)
7.A.4.2.1. Test Entry
Each test line is identified by letter names:
A, B, C..., Y. Whenever a test condition is referenced
in subsequent test or command lines, the letter name
of the test line is used.
USER LANGUAGE MANUAL
Language Specifications
7.A.4.2.1.
-215-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2030.1C01118 : CIA-RDP78-03952A000100070001-5
7.A.4.2.2. TAG Entry
This entry designates the tag or system value
being tested. It can be any of the following:
- any tag defined in the File Format Table of the
system data file. For the Document Index File
these will be the tags of the indexing language.
- an arbitrary tag which is defined in the current
job and which is referenced in an editing or
computational command (when files are not under
FFT control).
- A system generated value:
These include DEFINE n; TALLY n; ACCUM n.
See Sections 7.A.4.7.5., 7.A.4.8.1. and
7.A.4.8.2.
7.A.4.2.3. OP Entry
This entry can be either a numeric operator
or a relational operator.
A numeric operator specifies the least number of
values which must be found for the test to be satisfied.
Also, if more than this number is found, the condition
is met.
When a '1' appears with no value entry, the condi-
tion is counted as met if the tag specified in the tag
entry is found with any value. Figure 7.A-7 shows the use
of the OP numeric operator.
USER LANGUAGE MANUAL
Language Specifications
7.A.4.2.3.
-216-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/SKROA-RDP78-03952A000100070001-5
TEST
TAG
OP
VALUE(S)
A
PNO
1
2341
PEO
2
271 & 278
&
279
C
CNC
2
29 & 30
D
CNC
1
Figure 7.A-7
The Numeric OP Entry
Test line A - this simple test is satisified if
a PNO term with the value specified is found.
Test line B - this simple test is satisfied if
PNO terms with two of the three values (271, 278 or
279) are found.
Test line C - this simple test is satisfied if
CNC terms with both of the values are found.
Test line D - this simple test is satisifed if
any CNC term is found.
Several relational operators may be entered. The
values may be numeric or alphanumeric characters (see
Figure 7.A.8.).
USER LANGUAGE MANUAL
Language Specifications
7.A.4.2.3.
-217-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 200603Z132 : CIA-RDP78-03952A000100070001-5
TEST TAG OP VALUE (S)
A PDB > 310705
PDN < 290628
PDB X 280401 & 300101
PNO
POH N DIRECTOR
Figure 7.A-8
The Relational OP Entry
> (greater than) - The simple test condition A
is satisfied if a PDB term is found with a value greater
than July 5, 1931.
< (less than) - The simple test condition B is
satisfied if a PDN term is found with a value less than
June 28, 1929.
X (span) - The simple test condition C is satisfied
if a PDB term is found with a value greater than
April 1, 1928 and less than January 1, 1930. Values
must be separated by &. Only one & may appear on a line
with this operator.
USER LANGUAGE MANUAL
Language Specifications
7.A.4.2.3.
-218-
25X6
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/IMICTA-RDP78-03952A000100070001-5
The simple test condition D is satisfied if a PNO term
is found whose value is greater than
in the collating sequence of the IBM/360.
N (negation) - The simple test condition E is
satisfied if a POH term is found with a value other
than "DIRECTOR." (If no POE tag appears, or if it appears
with "DIRECTOR," the condition is not satisfied.) Thus,
this operator requires that (a) the tag in the tag entry
be present, (b) no value for that tag be equal to the
value entered.
7.A.4.2.4. Value(s) Entry
Each value for which a test is made is entered here
separated by &. One blank must appear on each side of
the &. Additional blanks are treated as part of the value.
Values can include the characters A-Z, 0-9 and CHIVE
special symbols as described herein.
Other types of test conditions can be specified by
appropriate special symbols in the value entry:
- Phrase coordination (*)
- Don't care characters (?, !)
- Tag occurrence (#)
USER LANGUAGE MANUAL
Language Specifications
7.A.4.2.4.
-219-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
25X6
Approved For Release 2008B1394108 : CIA-RDP78-03952A000100070001-5
- Indirect value specification ($ FILE)
Figure 7.A-19 in Section 7.A.4.3.5. shows the
permissable combinations of the special value symbols
with various operator field entries.
7.A.4.2.4.1. Phrase Coordination
Phrase coordination occurs whenever the values
specified must be found or satisfied within the same
phrase rather than merely within an entire record. The
values must be preceded by an 11" (see line A, Figure 7.A-9).
When the special symbol, '*', is used in a simple
test, the test line should specify that more than one
value be matched. Lines B and C are not meaningful uses
of the "I.
TEST
TAG
OP VALUE(S)
A
PNO
2 * 2174
&
2175
PNO
1 *2174
PNO
1 * 2174
&
2175
PNO
2 * 2175
&
2176
& 2177
Figure 7.A-9
Value Entries with Phrase Coordination
USER LANGUAGE MANUAL
Language Specifications
7.A.4.2.4.1.
-220-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/02ECCIA-RDP78-03952A000100070001-5
7.A.4.2.4.2. Don't Care Characters
There are two "don't care" characters available:
says to ignore the single character
appearing in the corresponding position
in the value of the record term currently
being examined.
! says to ignore a variable number of
characters in the corresponding position
in the value of a record term currently
being examined. This ignoring
the character following the !
encountered in the record term
stops when
is
value.
TEST
TAG
OP
VALUE(S)
A
PNO
1
B
PDB
1
C
PNO
1
D
LAC
1
E
LAC
1
1??3!
F
PDT
X
??0564
&
??0665
Figure 7.A-10
Value Entries with DON'T CARE Characters
The test conditions in Figure 7.A-10 are satisfied
i.e., they are true, as follows:
USER LANGUAGE MANUAL
Language Specifications
7.A.4.2.4.2.
-221--
Approved For Release 2000/05/08 : ClA:RDP78-03952A000100070001-5
SECRE
25X6
25X6
25X6
25X6
Approved For Release 20aMt8 : CIA-RDP78-03952A000100070001-5
Test condition A--All characters, except the character
in the position corresponding to the ?, match exactly.
Test condition B--All characters, except the four
characters whose positions correspond with the four ?'s
match exactly. There must be exactly six characters in
the value being tested.
Test condition C--The first eight characters must
match An exact match on the characters which
folio is not required. Any number of
characters may follow111111111
Test condition D--The first three characters must match
Any single character may folloul. and there
must be at least one such character. This character
may be followed by any number of character(s), or by no
characters at all. The last four characters must match
PING exactly.
Test condition E--The first character must be a
numeric I followed by any two (but only two) characters.
The fourth character must be a numeric 3. The 3 may be
followed by any number of any characters or by no
characters at all.
USER LANGUAGE MANUAL
Language Specifications
7.A.4.2.4.2.
-222-
25X6
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08MA-RDP78-03952A000100070001-5
Test condition F--The values for the PDT tested must be
such that (a) the last four characters are greater
than 0564, and (b) the last four characters of the
same value are less than 0665.
The don't care characters may not have an alpha-
numeric character on both sides when used with < , >,
and X operators.
7.A.4.2.4.3. Tag Occurrence Symbol
The actual number of occurrences of the tag speci-
fled in the TAG entry must satisfy the relational
condition specified in the OP entry with regard to
the number "n" following the # symbol.
Examples of the tag occurrence symbol, "#" are
shown in Figure
7.A-11.
TEST
TAG
OP
VALUE(S)
A
PNO
>
#1
B
PNO
<
#15
C
PNO
1
#4
D
PNO
N
#3
E
PNO
1
#3
Figure 7.A-11
Value Entries with Tag Occurrence Symbol
-223-
USER LANGUAGE MANUAL
Language Specifications
7.A.4.2.4.3.
Approved For Release 2000/05/0VerRDP78-03952A000100070001-5
Approved For Release 206608?8 : CIA-RDP78-03952A000100070001-5
- Test condition A--PNO must occur more than
once.
- Test condition B--PNO must occur less than
15 times.
- Test condition C--PNO must occur exactly four
times.
- Test condition D--PNO must occur any number
of times other than three times
- Test condition E--PNO must occur three times
in the same phrase.
7.A.4.2.4.4. $ FILE Indirect Specification of Values
The special symbols, $ FILE, specify that the characters
immediately following these constitute the name of a file.
The actual values will be obtained from the records of this
file.
Examples of $ FILE (name) are shown in Figure 7.A-13.
This feature is used in enriched searching and is more
fully explained in Section 7.A.5.1.
TEST TAG OP VALUE(S)
A PNO 1 $ FILE OUT 1
PDB $ FILE OUT 2
Figure 7.A-13
Value Entry with $ FILE (name)
USER LANGUAGE MANUAL
Language Specifications
7.A.4.2.4.4.
-224-
Approved For Release 20E1M8 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/0ftcW-RDP78-03952A000100070001-5
Test condition A--is satisfied if the record under
consideration has at least one value equal to any of the
PNO terms in File OUT 1.
Test condition B--is satisfied if the record under
consideration has a PDB value greater than any one of the
PDB values in File OUT 2.
Any file referenced by $ FILE (name) in the VALUE
field must exist in the system or must have been previously
generated for the current transaction.
7.A.4.3. Complex Test Lines
A complex test line specifies a test condition made up
of simple test conditions. It has no TAG entry. Value
entries are always alphabetic names (A,B,...) of previously
specified test lines.
7.A.4.3.1. Example of a Complex Test Line
Locate all records containing both
25X6
25X6 with an LAC of 2314 or 2715.
USER LANGUAGE MANUAL
Language Specifications
7.A.4.3.1.
-225-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2oomtict : CIA-RDP78-03952A000100070001-5
TEST
TAG
OP
VALUE(S)
A
PNO
2
25X6
B
LAC
1
2314 & 2715
C
2
A & B
Figure 7.A-14
Test Line (C) Made Up of Test Lines (A,B)
Test Line C is a complex test statement made up
of the simple test statements on lines A and B. The
'2' in the OP field indicates that both test A and test B
must be true to satisfy test line C.
The Boolean expression equivalent to Test line C
is: AND
(LAC 2314 or LAC 2715).
7.A.4.3.2. Phrase Coordination in a Complex Test Line
The phrase coordination symbol '*' may also be
used as a value entry preceding the test line names.
As an example: identify the Organization Head for the
. The organization
type number is 6700. Figure 7.A-15. shows the necessary
entries.
USER LANGUAGE MANUAL
Language Specifications
7.A.4.3.2.
-226-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/58CRER-RDP78-03952A000100070001-5
TEST TAG OP VALUES
A LAC 1
B ONO 1
C OTF 1 6700
D POH 1
4 *A&B&C&D
25X6
Figure 7.A-15
Phrase Coordination in a COMPLEX Test Line
Since the example calls for the occurrence of any
POH in a phrase with conditions A, B, C, no value for
line D is given and the numeric operator is a '1.'
The condition defined on Line E, preceded by an * requires
that the 4 conditions (A, B, C and D) be met within a phrase.
No complex test line can contain an * if it
references a preceding test line which also contains an
(see Figure 7.A-16.).
USER LANGUAGE MANUAL
Language Specifications
7.A.4.3.2.
-227-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 248M8 : CIA-RDP78-03952A000100070001-5
TEST
TAG
OP
VALUES
Q
PEO
2
*27 & 28 & 29
PERMISSABLE
TEST LINES
TEST
TAG
OP
VALUES
(4 & C)
A
PEO
2
27 & 28
B
PNO
1
2784, 2794
* A & B
TEST
TAG
OP
VALUE ($)
ERRONEOUS
A
PEO
2
*27 & 8 & 9
TEST LINE (C)
B
PNO
2
2784, 2794
C
2
* A & B
Figure 7.A-16
The Uses of 1" for Phrase Coordination
7.A.4.3.3. M of N Logic in a Complex Test Condition
Often, it may be useful to ask that at least two
of three; or, two of four of the test conditions named
be satisfed in a phrase or record.
USER LANGUAGE MANUAL
-228- Language Specifications
7.A.4.3.3.
Approved For Release 20@kftfa8 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/03KM-RDP78-03952A000100070001-5
As an example the user wishes to identify a
conference attended by at least two, and possibly more
than two of the PNO's listed on test line B. The
conference date is August 31, 1964. Figure 7.A-17
shows the necessary test conditions.
TEST TAG OP VALUE(S)
A CDY 1 640831
PNO 2 2314 & 3124 & 4675 & 7124
CNC 1
3 *A&B&C
Figure 7.A-17
Test Lines with M of N Logic
The entry in the OP column for line B is 2, indicating
that the condition is met if at least two (and 2ossib1y
more) of the PNO values appear.
7.A.4.3.4. Span Operator in a Complex Test Condition
The following example shows how a test line with
the span operator is referenced in a complex test
condition. Determine the phrases mentioning all PNO's who
attended PEO 23, PEO 26, and PEO 28 during the first six
months of the year 1932.
USER LANGUAGE MANUAL
Language Specifications
-229- 7.A.4.3.4.
Approved For Release 2000/05/08ac ? ClAiRDP78-03952A000100070001-5
RE
Approved For Release 200R6M : CIA-RDP78-03952A000100070001-5
TEST TAG OP VALUE
A PEO 1 23, 26, 28
PNO 1
PDE X 311231 & 320701
3 *A&B&D
Figure 7.A-18
Example of Use of X Operator
7.A.4.3.5. Summary of Rules for Test Lines
The following matrix shows the permissible
combinations of special symbols with Operator field
entries.
Operators:
Symbols
X
N
1
Other number
&
no
no
yes
no
yes
yes
?
opt
opt
opt
opt
opt
opt
!
no
no
no
opt
opt
opt
$
opt
opt
no
opt
opt
no
#
opt
opt
opt
no
opt
no
no
no
no
no
no
yes
Figure
7.A-19
Summary of Test Line Rules
USER LANGUAGE MANUAL
-230- Language Specifications
Approved For Release 2g2g41508 : CIA:FtD167/3*-83952A000100070001-5
Approved For Release 2000/057MRTIA-RDP78-03952A000100070001-5
7.A.4.4. Basic Retrieval Commands: COPY, PRINT
Most retrieval transactions will require the use of
only the two commands--COPY and PRINT.
COMMAND LIMIT CRITERIA FILE DATA
R P T IN OUT
COPY
(Opt) v v v
PRINT (Opt)
Figure 7.A-20
COPY and PRINT Command Formats
7.A.4.4.1. COPY Command
The COPY command copies those records, phrases, or
terms which meet the search criteria. Section 7.A.4.5.
discusses the search CRITERIA entries in more detail
and should be read in conjunction with the following
description of the COPY and PRINT commands.
Rules:
1. Input and output files must be named.
2. Each record of the input file is examined to
determine whether or not it contains a data element
that meets the search criteria.
3. Record, phrase, and term criteria entries are
required as follows:
- To copy a term: record, phrase, and term (R, P, T)
criteria entries are required.
USER LANGUAGE MANUAL
Language Specifications
Approved For Release 2000/05/1E. -RDP78-605?AbOb100070001-5
Approved For Release 2000NOBF08 : CIA-RDP78-03952A000100070001-5
- To copy a phrase: phrase and term (P, T) criteria
entries are required.
- To copy a record: only the record (R) criterion
entry is required.
4. The LIMIT entry is optional
5. No DATA entry is used.
The COPY command does not retrieve data into a
temporary work area for editing or for computation. To
perform post-retrieval operations other than printing,
the EXTRACT command should be used.
The COPY command copies the data elements (records,
phrases, or terms) specified by appropriate use of the
Search Criteria entries. Each data element (R, P, or T)
of the named input file that is specified as a Criteria
entry is examined to determine that it meets the
condition named by the Search Criteria entries. If all
test conditions, so named are met for each data element
specified, the data element of the lowest level specified,
i.e., record (R), phrase (P), or term (T) is copied in
the named output file as a record. Figure 7.A-21 shows
examples of Search Criteria at the three levels.
USER LANGUAGE MANUAL
Language Specifications
7.A.4.4.1.
-232-
Approved For Release 2ORM/108 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/0fEccitit-RDP78-03952A000100070001-5
Figure 7.A-21
COPY and PRINT Command Lines
Record Level Search Criteria
COMMAND LIMIT CRITERIA FILE
R P T IN OUT
DATA
COPY A DIF OUT 1
PRINT OUT 1 RCD
EXAMPLE A
Phrase Level Search Criteria
COMMAND LIMIT CRITERIA FILE DATA
R P T IN OUT
COPY
PRINT
50 A B DIF OUT 1
OUT 1 PHR PNO
EXAMPLE B
Term Level Search Criteria
COMMAND LIMIT CRITERIA FILE
R P T IN OUT
DATA
COPY
PRINT
Z B C DIF OUT
OUT TER PEO & PNO
EXAMPLE C
-233-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 200980/08 : CIA-RDP78-03952A000100070001-5
EXAMPLE A--COPY records satisfying the
condition on test line A.
EXAMPLE B--COPY phrases satisfying the condition
on test line B from records satisfying condition on
test line A. The LIMIT entry limits the retrieval
to a maximum of 50 hit phrases for output.
EXAMPLE C--COPY terms satisfying condition on
test line C in those phrases satisfying condition on
test line B. The Z entered for record criteria entry
(R) indicates that phrases and terms from all records
are examined for the test conditions in the phrase (P)
and term (T) Criteria entries. (A test line could also
be referenced in the R field if it is needed.)
7.A.4.4.2.PRINT COMMAND
The PRINT command prints a named Input File. PRINT
designates a standard output format in the Output File
entry. This command, in conjunction with the COPY
command, constitute the basic retrieval command set.
Rules:
1. An Input File name must be specified.
2. The standard output format must be named as an
OUT FILE entry.
USER LANGUAGE MANUAL
Language Specifications
7.A.4.4.2.
-234-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
tor
Approved For Release 2000/05/01ECIMA-RDP78-03952A000100070001-5
3. Sort keys can appear as DATA entries. Where
multiple sort keys are specified, each key is separated
by an &.
4. If there is more than one occurrence of a sort
key in a record, that record is repeated in the output
listing for each occurrence of individual sort key
values.
5. The DATA field can contain additional parameters
required for the named output format.
As an example, COPY and PRINT all records mentioning
25X6 in the same phrase. The
necessary entries are shown in Figure 7.A-22 and 7.A-23.
TEST TAG OP VALUE(S)
A PNO 2
LAC 1 2314 & 2715
* A & B
Figure 7.A-22 - Test Line
COMMAND LIMIT CRITERIA FILES DATA
R P T IN OUT
COPY C DIF OUT 1
PRINT OUT 1 RCD
Figure 7.A-23
1.10 COMMAND lines for COPY and PRINT
USER LANGUAGE MANUAL
sir -235- Language Specifications
Approved For Release 2000/05/08 iRODP78-03952A0.0100070001-5
Approved For Release 2066kW8 : CIA-RDP78-03952A000100070001-5
Figure 7.A-21 also shows several uses of the PRINT
command. The first example shows the use of the PRINT
command to print hit records. The named input file is
OUT 1. The OUT FILE entry RCD, names a standard report
format used for printing records. The second example
shows the use of the PRINT command to print phrases.
The OUT FILE entry,PHR, names a standard report format
to print phrases. The DATA entry contains the sort
key PNO. Major and intermediate sort keys, PEO and PNO,
are shown in the third PRINT example of Figure 7.A-21.
7.A.4.5. Search CRITERIA and DEFINE Command
Figure 7.A-24 shows additional search criteria
entries
element
used in a COPY
desired.
command depending
upon
the data
COMMAND
LIMIT
CRITERIA FILES
DATA
R P T IN
OUT
#
1
COPY
A bIF
OUT
1
#
2
COPY
A B DIF
OUT
2
#
3
COPY
z B C DIF
OUT
3
Figure 7.A-24
CRITERIA entries for Various Data Elements
Line # 1--COPY all records satisfying test condition
on test line A.
-236-
USER LANGUAGE MANUAL
Language Specifications
7.A.4.5.
Approved For Release 21gappio8 : CIA-RDP78-03952A000100070001-5
25X6
25X6
Approved For Release 2000/05/0/tcM-RDP78-03952A000100070001-5
Line # 2--COPY all phrases satisfying conditions on
test line B from records satisfying conditions on test
line A.
Line # 3--COPY all terms satisfying the condition
defined on test line C in all phrases satisfying the
condition on test line B. The Z entry in the record
level (R) entry indicates that the terms from all records
satisfying both the term and phrase level criteria are
to be copied.
7.A.4.5.1. Example of Complex Search Criteria
"Copy all phrases with references to
if they occur in records referring to both of these
personalities and a PAF reception somewhere in the
between
July 1955 and June 1964." The first figure, Figure 7.A-25,
shows Lie CRITERIA entries necessary to copy the phrases
on the two personalities. The CRITERIA phrase (P)
entry is C.
Notice the complex test lines which refer to simple
test lines: line C refers to A and B; line F refers to D
and E. Notice, also, that line I is a complex test line
that refers to both simple test lines, G, H, and other
previously defined complex lines, C, F.
USER LANGUAGE MANUAL
mew Language Specifications
7.A.4.5.1.
-237-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2088111a/08 : CIA-RDP78-03952A000100070001-5
Figure 7.A-25
Use of CRITERIA Entry
TEST TAG OP VALUE(S)
A
PNO 1
PNO 1
2
LAC 1
LAC 1
F 1 D&E
G HDR X 5507?? & 6406??
H PAF 1 !RECEPTION!
I 4 C&F&G&H
COMMAND LIMIT CRITERIA FILES DATA
R. P T IN OUT
COPY
PRINT
I C
DIF OUT 1
OUT 1 PHR
-238-
25X6
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/68GRZIA-RDP78-03952A000100070001-5
COMMAND LIMIT CRITERIA FILES DATA
R P T IN OUT
COPY I H DIF OUT
PRINT
OUT 1 PHR
Figure 7.A-26
Alternative use of CRITERIA Entry
Figure 7.A-26, shows the search CRITERIA entries
for a problem similar to the example given above. However,
in this example there is a change in the phrases to be
copied. Previously, phrases containing the personalities
named were copied. Here, the phrases containing a PAF
of "reception" are to be copied. Thus, the phrase (P)
entry is H. It is important to realize that in both
examples, although the phrases copied differ, they are
copied from records that satisfy identical test conditions.
Accordingly, the record CRITERIA entry for both examples
is I. (The test lines for both examples are identical.)
7.N.4.5.2. Define (m)
A data element (record or phrase) specified with
appropriate CRITERIA entries can be uniquely identified
by use of the DEFINE command.
USER LANGUAGE MANUAL
Language Specifications
7.A.4.5.2.
-239-
Approved For Release 2000/05/CTEM-RDP78-03952A000100070001-5
Approved For Release 209garEt08 : CIA-RDP78-03952A000100070001-5
COMMAND LIMIT CRITERIA FILE DATA
R P T IN OUT
DEFINE 1 Z E
EXTRACT
APPEND Z G
WRITEP
File 1
File 1
OUT 1
$ DEFINE 1
Figure 7.A-26A
DEFINE Command Line
The DEFINE command uniquely defines the data element
specified in the CRITERIA entries for this command.
Rules:
1. CRITERIA entries must be specified as in EXTRACT
and COPY commands.
2. The LIMIT entry is optional.
3. An Input File name must be specified.
4. "DEFINE" in the COMMAND entry is the name being
assigned to the data element specified in the CRITERIA
entries.
Subsequent references to this data element appear as
DATA entries "DEFINE." In any such reference, the "DEFINE"
name must be used.
5. Each DEFINE named must be unique for a retrieval
transaction.
USER LANGUAGE MANUAL
Language Specifications
-240- 7.A.4.5.2.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05WMA-RDP78-03952A000100070001-5
6. A DEFINE name may be referenced as a Tag field
entry in a test line by its assigned name.
Explanation
Those phrases which meet the condition specified on
test line E and which occur in the named input file are
named DEFINE I (see Figure 7.A-26A). Since the APPEND
command has as its DATA operand "$DEF1NE 1," the phrases
meeting the search CRITERIA for the DEFINE 1 command
are appended to records extracted by the succeeding
EXTRACT command in the example.
7.A.4.6. EXTRACT and WRITE
The COPY command copies retrieved data elements
directly into the named output file. When it is
necessary to edit or modify hit elements, the EXTRACT-
WRITE combination must be used.
7.A.4.6.1. EXTRACT
Data elements specified in the CRITERIA entries
are extracted into a work area for access and manipulation
by subsequent commands.
USER LANGUAGE MANUAL
Language Specifications
7.A.4.6.1.
-241-
Approved For Release 2000/05/11LCRIerRDP78-03952A000100070001-5
Approved For Release 2065%568 : CIA-RDP78-03952A000100070001-5
COMMAND LIMIT CRITERIA FILE
R P T IN OUT
DATA
TALLY 1 v v
EXTRACT (opt) v v v DIF
APPEND v v $TALLY1
WRITER OUT 2
PRINT OUT 2 PHR sort order key
Figure 7.A-27
Example of EXTRACT Command
Rules:
1. All search CRITERIA entries are required as in the
COPY command.
2. An Input File must be named. No output file name
is used.
3. The EXTRACT command may be followed only by APPEND,
DELETE, REPLACE, MERGEP, WRITEP, or WRITER commands.
4. All EXTRACT commands within the scope of the same
WRITER and WRITEP commands (see these commands below) must
have the same input file names.
5. The LIMIT entry is optional.
Explanation
Phrases or terms which meet the search CRITERIA are
extracted from a particular input record into a temporary
work area as a single record. Hit records are similarly
42-
USER LANGUAGE MANUAL
-2
Approved For Release 2Nt4k5/08 : CIA-Orn93/9-VN010f011ittefff01-5
Approved For Release 2000/0SIOGRECA-RD1278-03952A000100070001-5
copied into the work area for subsequent editing, computation,
etc.
A succeeding WRITER or WRITEP command is necessary to
copy what has been extracted into a named output file.
7.A.4.6.2. WRITER and WRITEP Commands
These commands write the contents of a work area
containing extracted data elements into a named output
file. The WRITER command writes out the contents of the
work area as an individual record. The WRITEP command
writes each phrase of a record from the work area as
an individual record.
Rules: Figure 7.A-28 shows examples of proper and
improper application of the following rules:
- An output file must be named by the WRITE command.
- Each WRITE command must be preceded by at least
one EXTRACT command.
- No other commands may intervene between a WRITE
and its preceding EXTRACT except: APPEND, DELETE,
REPLACE, and MERGEP.
- The command sequence EXTRACT. . ?EXTRACT.
WRITE. . .WRITE is not permissible.
USER LANGUAGE MANUAL
Language Specifications
7.A.4.6.2.
-243-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2006M0111: CIA-RDP78-03952A000100070001-5
EXTRACT
permissible APPEND
WRITER
EXTRACT
APPEND
permissible WRITER
APPEND
WRITER
EXTRACT
APPEND
not EXTRACT
permissible WRITER
APPEND
WRITER
EXTRACT
not APPEND
permissible MERGER
WRITER
Figure 7.A-28
Examples of WRITER COMMAND
Explanation
The WRITER command writes out the contents of the
work area as records. Where the CRITERIA entry for the
preceding EXTRACT specifies terms, all "hit" terms from
each input record are written as a single record in the
named output file. Where the CRITERIA entry for the
preceding EXTRACT specifies phrases, all hit phrases from
each input record are written as a single record in the
USER LANGUAGE MANUAL
Language Specifications
7.A.4.6.2.
-244-
Approved For Release viEropitos : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/059111CIMA-RDP78-03952A000100070001-5
named output file. Where the CRITERIA entry for the
preceding EXTRACT command specifies records, each hit record
of the input file is written in the named output file.
The WRITEP command writes out each phrase in the work
area as a separate record.
7.A.4.7. EDITING Commands: APPEND, DELETE, REPLACE
The data EXTRACTed into a work area can be edited
or modified by appending, deleting, replacing data
elements extracted into this area.
7.A.4.7.1. APPEND Command
The APPEND command appends data which appears as a
DATA entry to the data element in the work area which
meets the search CRITERIA.
COMMAND
LIMIT
CRITERIA
FILE
R P T IN OUT
DATA
EXTRACT
APPEND
WRITE
EXTRACT
2 APPEND
3 APPEND
WRITE
Z D
A
Z B
(GRP) ONE
(PNO) 2314 (PEO)
$Tally 10
Figure 7.A-29
Examples of APPEND Command
USER LANGUAGE MANUAL
Approved For Release 2000/05/0-82:461A-RDP78-03952A00040001900414cations
SECRET 7.A.4.7.1.
Approved For Release 20813231E/D8 : CIA-RDP78-03952A000100070001-5
Rules:
1. Data is always appended to the record which has
been extracted into the work area by the EXTRACT command
immediately preceding this command. APPEND must
be preceded by an EXTRACT command, and it must be within
the scope of a WRITE command.
2. Phrases or terms can be appended to hit records
and terms appended to hit phrases. Thus, the search
CRITERIA are record (R) or record and phrase, (R) and (P),
entries only.
3. Data to be appended can be:
- Any term defined in the File Format Table. If
the user is not operating under FFT control, then
any user specified term (see 1 and 2, Figure 7.A.29).
- A $TALLY, $ COMPUTE $ ACCUM, or $ DEFINE.
(see 3, Figure 7.A-29).
4. The LIMIT entry is optional.
Explanation
Data is added to those elements in the work area record
which meet the search CRITERIA. Terms can be appended
to phrases; or, phrases or terms can be appended to records
(see Figures 7.A-30 and 7.A.-31).
USER LANGUAGE MANUAL
Language Specifications
7.A.4.7.1.
-246-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/0kEttEfIA-RDP78-03952A000100070001-5
Record HEADER PHRASE
Control
Phrase 1
Phrase 2
Phrase 3
Phrase 4
New Phrase
Figure 7.A-30
New Phrase
Phrase Appended to a Record
Record
ControlTTTT
TTTT
TTTT
New
Term T
Figure 7.A-31
Term Appended to Phrase
Terms can be appended to phrases which meet the
conditions of a referenced test line as in 1, Figure 7.A-29.
Phrases can be appended to records which meet the
conditions of a reference test line as in 2, Figure 7.A-29
Data elements indirectly referenced may be appended
as in 3, Figure 7.A-29. Here the result of a TALLY command,
named TALLY 10, will be appended to phrases satisfying the
condition on test line B for all records.
A complete record
cannot be
appended to the file with
USER LANGUAGE MANUAL
ApproviefisFociRelease 2000/051024:7CIA-RDP78Aag5244001-14013DY080tlifins
SECRET
Approved For Release 20033/01838 : CIA-RDP78-03952A000100070001-5
7.A.4.7.2. DELETE Command
The DELETE command deletes data elements which meet
the specified Search Criteria from the record in the
temporary work area.
COMMAND LIMIT CRITERIA
R P ,T
I DELETE Z A C
2 DELETE Z B
3 DELETE A
Figure 7.A-32
Examples of DELETE Command Lines
Rules:
1. The DELETE command operates on data extracted into
a work area except for the record level. It must be
preceded by an EXTRACT command, and it must be within the
scope of a WRITE command.
2. Terms, phrases, or records can be deleted as in
Figure 7.A-32, 1, 2, and 3 respectively.
3. The LIMIT entry is optional.
Explanation
Data which meets the search CRITERIA are deleted
from the data element specified (see Figure 7.A-32, 1,
2, 3).
-248-
USER LANGUAGE MANUAL
Language Specifications
7.A.4.7.2.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/0$1131RECIA-RDP78-03952A000100070001-5
1 - Terms which satisfy test line C in phrases which
satisfy test line A are deleted from all records.
2 - Phrases which satisfy test line B are deleted
from all records.
3 - Records which satisfy test line A are deleted
from the file.
Only data elements in the work area record are
operated on by the DELETE command.
With FFT control OFF, terms with any tag (including
index language tags) can be deleted. With FFT control
ON, only tags defined in the File Format Table may be
deleted.
7.A.4.7.3. REPLACE Command
Data which appears as a DATA entry replaces the data
element specified in the CRITERIA entries.
COMMAND
LIMIT CRITERIA
R P T
DATA
I REPLACE
2 REPLACE
Z A C (PNO) 2145
B D (PED) 293 & PNO 2171
Figure 7.A-33
REPLACE Command Lines
USER LANGUAGE MANUAL
Language Specifications
-249-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 206e/t11811118 : CIA-RDP78-03952A000100070001-5
Rules:
1. The new data appears as a DATA entry.
2. Record (R), phrase (P) and term (T) CRITERIA
entries are required when terms are replaced.
3. Record (R) and phrase (P) entries are required
when phrases are replaced.
4. This command must be preceded by an EXTRACT, and
it must be within the scope of a WRITE command.
5. The LIMIT entry is optional for this command.
6. File control conventions apply as in the APPEND
and DELETE commands.
Explanation
The new data is replaced into the same location as
the term or phrase specified in the CRITERIA entries.
Data elements are replaced in Figure 7.A-33, 1 and 2,
as follows;
- Terms specified on test line C are replaced by
the term in the DATA entry. Replacements occur in
all phrases specified on test line A.
- Phrases specified on test line D are replaced
by the phrase appearing in the DATA entry.
Replacements occur in all records on test line B.
7.A.4.7.4. EXPLODE Command
The EXPLODE command copies phrases or terms onto
an output file as individual records.
-250-
USER LANGUAGE MANUAL
Language Specifications
7.A.4.7.4.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05gitizglA-RDP78-03952A000100070001-5
COMMAND LIMIT CRITERIA FILES
R P T
Phrase EXPLODE (Opt) v v v v
Explosion
Term EXPLODE (Opt) v v v v v
Explosion
Figure 7.A-34
EXPLODE Command Lines
Rules:
1. hit data elements are copied directly into a
named output file as individual records.
2. An input and an output file must be named.
3. The LIMIT entry is optional.
4. CRITERIA entries are required.
5. Only phrases and terms may be "exploded."
CRITERIA entries must appear under either (R),
(P), and (T) or under (R) and (P).
m0
Explanation
If phrases are EXPLODEd, those phrases meeting the
search CRITERIA from each record of the named input file
are copied in the named output file as individual records
as in Figure 7.A-35.
USER LANGUAGE MANUAL
Language Specifications
7.A.4.7.4.
-251-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 200CF/951118 : CIA-RDP78-03952A000100070001-5
INPUT RECORD EXPLODED RECORDS
Record
Control
Header Phrase,
(HIT)
Phrase 1
Phrase
2
Phrase
3
(HIT)
Phrase
4
Phrase
5
(HIT)
Phrase
6
Figure 7.A-35
Output Record 1
ThOl&Phrase
1
Output Record
2
Old Phrase
4
Output Record
3
Old Phrase 6
Exploded Hit Records
If term, (R), (P)_and (T), CRITERIA are specified,
those terms meeting the search criteria in each input
record are copied as individual records in the named
output file as shown in Figure 7.A-36.
7.A.4.8. Computational Commands: TALLY, ACCUM, COMPUTE
These commands are used in performing computational
operations on CHIVE data elements:
TALLY - counts the number of times a data
element (record, phrase, term) satisfies the specified
CRITERIA test condition(s) within a specified logical
structure.
USER LANGUAGE MANUAL
Language Specifications
-252- 7.A.4.8.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0estattrRDP78-03952A000100070001-5
Figure T.A-36
EXPLODE Terms Command
Terms Satimfying
earth CRITERIA
in INPUT
RECORD
OLD
Term fro HEIR
OLD from Ph. 1
Term 3,
OLD from F.
Term 2
OLD
Term
from Ph.
-253-
EX DID OUTPUT
RECOR FROM
IWPUT REC. ABOVE
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2001MM: CIA-RDP78-03952A000100070001-5
ACCUM - accumulates a sum of the values of terms
which satisfy the specified CRITERIA test condition(s).
COMPUTE -performs arithmetic operations on CHIVE
data elements and on user defined values.
7.A.4.8.1. TALLY Command
The TALLY command increments a counter by one
for each occurrence of a specified term (or terms) within
a phrase; a specified phrase (or phrases) within a record;
or a specified record (or records) within a named input
file.
Tallying of: COMMAND LIMIT CRITERIA FILE
RP T
terms in records TALLY 1 (Opt) v v v v
phrases in records TALLY 2 it
v v v
records in file TALLY 3 IP v v
Figure 7.A-38
TALLY Command Lines
Rules and Explanation:
Search CRITERIA entries are required as follows:
- to TALLY occurrence of phrases which satisfy a
named test line within a record: record (R) and
phrase (P) CRITERIA entries are required.
- To TALLY occurrences of terms within a record:
record (R), phrase (P) and term (T) CRITERIA
entries are required.
-254-
USER LANGUAGE MANUAL
Language Specifications
7.A.4.8.1.
Approved For Release 2CERr8 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05$11tIRCIA-RDP78-03952A000100070001-5
- To TALLY records which satisfy a named test line
in the file: only a record (R) CRITERIA entry is
required. In this case, the TALLY is accumulative
throughout the job. Its contents are printed
automatically.
A unique arbitrary number is assigned to the TALLY
command as in Figure 7.A-38. The number of occurrences
of the data element specified in the criteria entry is
saved in a location with this name "TALLY." The
number of occurrences can be subsequently referenced by
use of the assigned name. "TALLY" names must be
assigned a number which is unique for the retrieval
transaction in which it appears.
Example: In all records of the named file, TALLY
the occurrences of phrases which satisfy the test condition
on test line C. Append the result of the TALLY operation
in each extracted record (see Figure 7.A-39).
Since the contents of TALLY 1 are being appended to
the record for later output, "TALLY 1" appears as a DATA
entry preceded
USER LANGUAGE MANUAL
Language Specifications
7.A.4.8.1.
-255-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2O 8J8 : CIA-RDP78-03952A000100070001-5
Test TAG OP VALUE(s)
A PNO
PEO
1
1
249
2
* A
&
B
COMMAND LIMIT CRITERIA FILE DATA
R P T IN OUT
TALLY 1 Z C
EXTRACT
APPEND
WRITE
PRINT
DIF
DIF
OUT 1
OUT 1 RCD
$TALLY 1
Figure 7.A-39
TALLY and APPEND Result for Output
The assigned name "TALLY n" can also be referenced
as a test line entry by its assigned test line name. It
must be defined as a tag in a test line for each trans-
action in which it is referenced
7.A.4.8.2. ACCUM n Command
This command accumulates a sum of values found in:
- Records
- The results of TALLY and COMPUTE operations
- The result of another ACCUM operation
-256-
USER LANGUAGE MANUAL
Approved For Release 200gRE1/3 : CIA-RDWMAtibiabbrg
25X6
Approved For Release 2000/05adREMA-RDP78-03952A000100070001-5
Rules:
1. Search CRITERIA entries are required always to the
term (T) level, i.e., record (R), phrase (P) and term (T)
CRITERIA entries are required.
2. Input file name must be entered.
3. Output file name is not entered.
4. Each ACCUM Command entry in a given retrieval trans-
action must be given a number "n" which is unique for
that transaction.
5. A $ACCUM n name can appear as a DATA entry in the
APPEND or REPLACE commands.
Explanation
The values of terms that satisfy search CRITERIA
entries are summed and stored in a location with the name
ACCUM n.
In Figure 7.A-40 sums are accumulated for each
input record of all NO0 values ("number of organizational
affiliations") that occur in a phrase with PNO
The values thus accumulated for each input record are
defined as ACCUM 1. This accumulated sum is printed
automatically.
Values can be also accumulated across the file
for TALLY and COMPUTE n values. They must be defined
as tags in test lines as in Figure 7.A-41. Here a TALLY
USER LANGUAGE MANUAL
Language Specifications
-257- 7.A.4.8.2.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
25X6
Approved For Release 20(14100E108 : CIA-RDP78-03952A000100070001-5
TEST TAG OP VALUE(S)
A PEO 1 77
B PNO 1 2374
C * A & B
TALLY 1 1
COMMAND LIMIT CRITERIA FILE DATA
R P T IN OUT
TALLY 1
ACCUM 1
Z C DIF
Z Z D DIF
Figure 7.A-41
Command Lines for Accumulating TALLY Result
is made in each input record of all phrases with PEO 77
and PNO 2374. (TALLY I command line.) ACCUM 1 produces
a sum for all values of TALLY 1 defined on Test Line D.
TALLY 1 contains the number of hit phrases for each
input record. The ACCUM 1 command line produces a total
of occurrences of these hit phrases for the entire file.
7.A.4.8.3. COMPUTE n Command
This command performs arithmetic operations on
numeric valued terms; values of TALLY and ACCUM counters;
values of other COMPUTE's; and user defined constants.
USER LANGUAGE MANUAL
Language Specifications
-258- 7.A.4.8.3.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/01A-RDP78-03952A000100070001-5
TEST TAG OP VALUE(s)
A NO0 1
B PNO 1
C * A & B
25X6
COMMAND LIMIT CRITERIA FILE DATA
R P T IN OUT
ACCUM 1
C A DIF
Figure 7.A-40
Command Lines for Accumulating Across a File
Rules and Explanation:
Each COMPUTE command of a retrieval transaction must
be assigned a number n which is unique for that transaction.
Operators and operands appear as DATA entries (see
Figure 7.A-42).
COMMAND LIMIT CRITERIA FILE
R P T IN OUT
DATA
COMPUTE 1
COMPUTE 2
COMPUTE 3
COMPUTE 4
COMPUTE 5
Tagl+Tag2-Tag3
ACCUM1*ACCUM2/TALLY7
Tagl*Tag2/TA-T3-Tag4
Tagl*Tag2-Tag3
Tag4/COMPUTE4
Figure 7.A-42
COMPUTE Command Lines
USER LANGUAGE MANUAL
Language Specifications
7.A.4.8.3.
-259-
Approved For Release 2000/05/08EORK-RDP78-03952A000100070001-5
Approved For Release 20V66g108 : CIA-RDP78-03952A000100070001-5
Operators are:
FIRST ORDER
SECOND ORDER
* multiply
+ add
The order in the
/ divide
- subtract
list above is the order in which
these operations will be performed for each COMPUTE
command. Multiply and divide are performed first; then
add and subtract. The operations performed in Figure
7.A-42 are performed by the system as follows:
COMPUTE 1: Add Tag 1 to Tag 2. Subtract Tag 3
from result.
COMPUTE 2: Multiply ACCUM 1 by ACCUM 2. Divide
product by TALLY 7.
COMPUTE 3: Tag 1 x Tag 2
Tag 3
- Tag 4
COMPUTE 4: Multiply Tag 1 by Tag 2. Subtract Tag 3
from product.
COMPUTE 5: Tag 4 divided by result of COMPUTE 4.
The COMPUTE command operates on one input record at
a time when its operands
are values of tags. In the
first command line of Figure 7.A-42, COMPUTE 1 is com-
puted and stored for each record. If operands cannot
be uniquely defined, the system regards this as an error
condition.
USER LANGUAGE MANUAL
Language Specifications
7.A.4.8.3.
-260-
Approved For Release 2411101135/08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/0-RDP78-03952A000100070001-5
Operands can be:
- Defined tags
- Values for counters, TALLY (n)
- Accumulated sums, ACCUM (n)
- Previously generated COMPUTE n results
- Defined constants
Each COMPUTE command must be assigned a number,n,
which is unique for the retrieval transaction.
The result is always stored in the name COMPUTE n,
and it must be referenced by this name.
A COMPUTE n name can be referenced as a DATA entry
in APPEND and REPLACE commands. It must be preceded
by $. (See Figure 7.A-43.)
COMMAND
APPEND
LIMIT CRITERIA
R P T.
Z C
COMPUTE 10
Figure 7.A-43
COMPUTE n Referenced in an APPEND Command Line
The result stored in COMPUTE 10 is appended to
each phrase which satisfies the CRITERIA entries for
the APPEND command.
USER LANGUAGE MANUAL
Language Specifications
7.A.4.8.3.
-261-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2stereesio8 : CIA-RDP78-03952A000100070001-5
7.A.5. INTER-RECORD AND INFERENTIAL LOGIC
This section describes those commands which perform
functions that require inter-record and inferential
logic.
7.A.5.1. Enriched Searching
Requests may entail inter-record logic against the
file. These requests often require two passes against
the entire file and the formulation of two integral query
requests by the user. The enriched query feature
of the CHIVE language enables the user to formulate
requests of this sort in a single retrieval transaction.
This class of requests retrieves all information
related to a logically determinable set of data elements
which cannot, however, be specified explicitly at request
time. For example, a request calls for biographical
information on personalities who have studied at the
since 1960. The
PNO's involved in this request cannot be specified at
request time. Satisfaction of the request normally
requires two passes to:
1. Determine the names (PNO's) of personalities
attending the INSTITUTE since 1960, and then
2. Copy all biographical information on each of the
PNO values which were generated in the first pass.
USER LANGUAGE MANUAL
Inter-record & Inferential Logic
7.A.5.1.
-262-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05?RROA-RDP78-03952A000100070001-5
The first search (see Figure 7.A-44) copies all
PNO's appearing in phrases mentioning the Institute
(394); the POE in these phrases will be later than
December 31, 1959 (1960).
This search will generate the values for all PNO's
to be used in the secondary search. The second search
then re-examines all records for phrases satisfying
the condition defined on test line F. Test line F is
satisfied if and only if a PNO value from the input (DIF)
file matches a PNO value in the file OUT 1. The special
characters, "$ FILE" are a requirement; they inform
the system that OUT 1 is the name of a file and not the
value for the tag entered on that line.
USER LANGUAGE MANUAL
Inter-record & Inferential Logic
7.A.5.1.
-263-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2020Ze5/108 : CIA-RDP78-03952A000100070001-5
TEST TAG OP VALUE(s)
A LAC 1 2147
B PEO 1 394
C PDE ) 591231
PNO 1
4 *A&B&C&D
PNO 1 $F1LE OUT 1
COMMAND LIMIT CRITERIA FILES DATA
R P T IN OUT
COPY Z E D DIF OUT 1
COPY Z F DIF OUT 2
PRINT OUT 2 PHR PNO
Figure 7.A-44
Enriched Querying
7.A.5.2. MERGER and MERGEP Commands
7.A.5.2.1. MERGER Command
The MERGER command causes all records, in a named
input file, which contain a common data element to be
co-located in a named output file. The user can request
a merge on a single tag or on multiple merge key tags.
USER LANGUAGE MANUAL
Inter-record & Inferential Logic
7.A.5.2.1.
Approved For Release 2800105/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05RtkRIA-RDP78-03952A000100070001-5
COMMAND LIMIT CRITERIA FILES DATA
R P T IN OUT
MERGER
_ _ _ X X PNO,
COMMAND LIMIT CRITERIA FILES DATA
R P T IN OUT
MERGER
- - - X X PNO, & CNC,
Figure 7.A-45
MERGER with Single and with Multiple Merge Keys
Rules:
1. Input and output files must be named.
2. A merge key must be specified as a DATA entry. All
records in which a common value for this key tag is found
co-located.
3. No CRITERIA entries appear with this command.
4. The MERGER command does not appear within the scope
of a WRITE command since this command initiates a file
operation.
Explanation
Those records with a common value for the merge
key are co-located in a named output file.
In Figure 7.A-46 records 1 and 2 contain the common
value, 20, for the merge key CNC. These records are
USER LANGUAGE MANUAL
Inter-record & Inferential Logic
7.A.5.2.1.
-265-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000aratr: CIA-RDP78-03952A000100070001-5
Figure 7.A-46
Input Records Before MERGER Command
RECORD
CONTROL
CNQ
1100
(Tag -
Value)
HEADER
(Tag -
Value)
(Tag -
Valu
Phrase 1
Phrase 2
Phrase n
RECORD
CONTROL
LAC 27
1 (,CNC 20 1
CSS 17
1
]
LAC 32
CNC 14
CSS 22
LAC 39
CNC 19
CSS 64
1
RECORD NO. 1
CNQ
1820
(Tag -
Value)
HEADER
(Tag -
Value)
(Tag
Valu]
Phrase 1
Phrase 2
Phrase n
LAC 29
CNC 26
CBS 16
1
I II
LAC 14
r -
t CNC 20)
-
CSS 41
?
LAC 13
CNC 23
Cm 3.6
__ ___,
1
RECORD NO. 2
-2 6 6-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/MMA-RDP78-03952A000100070001-5
merged into a single record as in Figure 7.A-47. The
new record control for the merged record is CNZ 20.
Records 1 and 3 will also be merged into a single record
since they both possess a CNC14. Each header of the old
records is retained with the new merged records.
If different values of the merge key are found in
exactly the same records, those records will be merged
into a logical record for each merge key value. Where
a value for the merge key occurs in only one record of
the named input file, that record is also copied in the
named output file. Thus in Record 3 of Figure 7.A-47,
if it is assumed that CNC 17 occurs only in that record
of the file, then Record 3 is copied in the named
output file as an individual record. (Also, records
I or 2 are copied in the output file as individual
records if any other CNC's occur only in either record.)
Where multiple keys are specified, those records
with common values for the combination of merge keys
are merged into a single record. In Figure 7.A-48,
records 1, 2, and 3 have the common pair CNC 1 and PNO9
and they are merged as in Figure 7.A-49.
For each combination of values which appears in
more than one record, a new record is created that is
made up of each record in which that combination of
common values appeared. When there is more than one
USER LANGUAGE MANUAL
Inter-record & Inferential Logic
7.A.5.2.1.
Approved For Release 2000/05/08-93174-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/85MT CIA-RDP78-03952A000100070001-5
Figure 7.A-47
MERGED RECORD with Common CNC 20
RECORD
CONTROL
CNQ
(TAG -
ell* -
(TAG ----
(TAG -
1920
VALUE)
VALUE)
HEADER
VALUE)
VALUE)
Phrase 1
Phrase 2
Phrase n
NEW RECORD
CONTROL
CNZ
20
CNC 17
CBS 17
?
EF177--1 CNC 14j CSS 22_1_ CJ
RECORD 3
HEADER (Old Record 1)
Vali
Value)
HEADER
CNC 20
(Old Record
CSS 17
2)
PHRASE 1
PHRASE 2
1AC.27
LAO 32
CNC 14
CSS 22
LAO 3
ONC 19
cm 64
PHRASE 3
LAC 29
CNC 26
CSS 16
CBS 41
PHRASE 4
PHRASE 5
_ PHRASE
LAC 14CN2O
LC 13
CNC 23
C 16
--------------
?
r--Eib 30ET 1k n
-2 6 8-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08SECKIRDP78-03952A000100070001-5
Figure 7.A-48 ?
Input Records with Common Values
for Multiple Merge Key PNO and CNC
CNC 1 . T-V
T-V
HEADER
TV
-
T-V
TV
NO 9
4
PAF 4
PCM 3
PED 7
PHRASE '
PNO 8 CDY 65
PHF 4
PRE 0
Pa' 29
PHRASE '
RECORD 1
CNC 1
T-V
T-V
HEADER
T-V
T-V
-INC 1
PNO 9
CSS 9
olo 6
ONP 6
PHRASE 1
PNO 8
CDY 65
PLP 64
POH 4
RMO 69
PHRASE 2
RECORD 2
CNC 1
TV
T-V
HEADER
T-V
TV
LAC 6
LAN 4
PNO 7
PNO 9
-,
CNC 1
PHRASE 1
PHRASE 2
LAC 7
CNC 4
CSS 22 PITO 8
cDY 65
RECORD 3
-269--
Approved For Release 2000/05/08 ; CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 200MEMI : CIA-RDP78-03952A000100070001-5
Figure 7.A-49
Merged Record with Common Multiple Merge Key
CNZ
1
PNZ
9
HEADER (Old Record 1)
HEADER (Old Record 2)
HEADER (Old Record 3)
CNC 1
PAF 4
PCM 3
PED 7
PHRASE 1
ENO 8
CLY 65
PHF 4
PHN 0
Pa 29
PHRASE 2
NC 1
PNO 9
CBS 9
OLP 6
ONP 6
PHRASE 3
NO 8
CD! 65
PLP 64
POH 4
MO 69
PHRASE 4
ZLAC 6
LAN 4
PNO 7
PNO 9
CNC 1
PHRASE 5
LAC 7
CNC 4
PNO 8
CD! 65
CDY 65
PHRASE 6
-270-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/0?tglEbA-RDP78-03952A000100070001-5
pair of common values in exactly the same combination
of records, those records are merged multiple times
into new records.
When a combination of multiple merge key values
appears uniquely in a record, that record is copied on
the named output file.
An example is given below to illustrate the use
of the MERGER command for "inferential searching."
Problem: The user seeks information on an unidentified
scientific conference which took place in MOSCOW on or
later than July 5, 1963. The names of several outstanding
scientific personalities who attended the conference are
known. (Again, the identity of the conference itself
is unknown.)
It is assumed that the results of an earlier request
transaction have shown that
not contained in any single
Thus the inter-record logic
the information desired is
record of the CHIVE files.
performed with the MERGER
command is required for this transaction.
-271-
USER LANGUAGE MANUAL
Inter-record & Inferential Logic
7.A.5.2.1.
Approved For Release 2000/o5igkR911A-RDP78-03952A000100070001-5
Approved For Release 2000/03101SETCIA-RDP78-03952A000100070001-5
TEST TAG OP VALUE(s)
A LAC 1 9993, MOSCOW
CDY > 630705
CNC 1
PNO 1 2344 & 2344 & 2346 & 2347 & 2348
4 *A &B&C& D
PNO 3 2344 & 2345 & 2346 & 2347 & 2348
COMMAND LIMIT CRITERIA FILES DATA
R P T IN OUT
COPY E DIF OUT 1
MERGER OUT 1 OUT 2 CNC
COPY F OUT 2 OUT 3
PRINT OUT 3 RCD
Figure 7.A-50
Test and Command Lines for Inferential MERGER Logic
Each command in Figure 7.A-50 is explained as
follows:
COPY - All records are copied which mention a
conference in Moscow later than July 5, 1963 and at
least one of the known personalities. The input file
is the Document Index Records File (DIF). The output
file is named OUT 1.
USER LANGUAGE MANUAL
Inter-record & Inferential Logic
7.A.5.2.1.
Approved For Release 2000M88 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/0510eRISCIA-RDP78-03952A000100070001-5
MERGER - Using the OUT 1 file as input, all records
with a common CNC are merged.
All records from file OUT 1 with common values for
the CNC tag have been merged into single logical records
and put into file OUT 2. It is now possible to search
each of these merged records in file OUT 2 for mention
of the five known personalities.
COPY - Any record in file OUT 2 mentioning three or
more of the five known personalities is copied on file
OUT 3. Ideally, the search will retrieve at least one
record mentioning all of the known personalities. And, if
each of the personalities has been indexed with the CNC
that is sought, at least one record mentioning all of
them and that CNC will be retrieved. The numeric operator
'3' is used in the event that two of the personalities
might not have been indexed in any document index records
with the CNC sought.
All hit records are then PRINTed in a standard
format for records.
7.A.5.2.2. MERGEP Command
All phrases within a record with a common value for.
the merge key are combined or merged into a single new
USER LANGUAGE MANUAL
Inter-record & Inferential Logic
7.A.5.2.2.
-273-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2001MN : CIA-RDP78-03952A000100070001-5
phrase. Phrases can be merged on a single merge key or
on multiple merge keys. (Examples of both cases are
given in Figure 7.A-51.)
COMMAND LIMIT CRITERIA FILE
R P T IN OUT
DATA
MERGEP _ _ _ PNO
COMMAND LIMIT CRITERIA FILE
R P T IN OUT
DATA
MERGEP
_ _ _
PNO & PEO
Figure 7.A-51
MERGEP with Single and Multiple Merge Keys
Rules:
1. Input and output files are not named.
2. A merge key must be specified as a DATA entry. All
phrases in which a common value for this key are found
are merged into a single phrase. Multiple merge keys
may be specified. They must be separated by "&".
3. No LIMIT or CRITERIA entries are used.
4. This command must be preceded by an EXTRACT command,
and it must be within the scope of a WRITE command. The
command operates on data which is retrieved by an extract
command, i.e., on data which is in the work area.
USER LANGUAGE MANUAL
Inter-record & Inferential Logic
7.A.5.2.2.
Approved For Release 200=6108 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05$01M1A-RDP78-03952A000100070001-5
Explanation
All phrases within a record with common values for
the merge key are merged into a single new phrase.
RECORD
CONTROL HEADER PHRASE
PNO 1 PEO 7 PNO 2 PEO 4 PNO 3 Phr.1
PNO 1 PEO 8 PNO 2 PEO 9 PNO 3 Phr.2
PNO 1 PEO 10 PNO 4 PEO 6 PNO 5 Phr.3
PNO 6 PEO 1 PNO - PEO - 1)&0 - Phr.4
PNO 7 PEO 14 PNO - PEO PEO - Phr.5
PNO - PEO - PNO'- PNO - PEO Phr.6
PNO - PEO - PNO - PNO - PEO Phr.7
Figure 7.A-52
Phrases with Common Values for Merge Keys
In figure 7.A-53, Phrases 1, 2, and 3 have a common
PNO value (1). Phrases 1 and 2 also have common PNO
values (2) and (3).
If more than one merge key value is common to exactly
the same phrases, those phrases will be merged into a new
single phrase only once. This convention is different
from the MERGER convention for different merge key values
occurring in exactly the same records. The matrix in
USER LANGUAGE MANUAL
Inter-record & Inferential Logic
7.A.5.2.2.
-275-
Approved For Release 2000/05/08 : QIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2006M61 : CIA-RDP78-03952A000100070001-5
LIWt
Figure 7.A-53 shows that PNO values 2 and 3 are common
to exactly the same phrases, namely phrases 1 and 2.
Thus, only one new phrase is created from phrases 2 and
3 below, even though both values 2 and 3 are found in these
phrases.
Phi Ph 2 Ph 3 Ph 4 Ph 5
1
2
3
4
5
6
7
PNO Values
Figure 7.A-53
Matrix of VALUES for Phrases 1-5
Every unique combination of phrases containing a
common value is merged into a single, new phrase.
The application of the above rules to the record
in Figure 7.A-53 results in the following phrase merging
operations:
- Phrases 1, 2, and 3 into a single phrase for
PNO 1, as shown in Figure 7.A-54.
USER LANGUAGE MANUAL
Inter-record & Inferential Logic
-276-
Approved For Release 200MVA8 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/0?EMA-RDP78-03952A000100070001-5
- Phrases 1 and 2 into a single phrase for PNO 2
since it occurs in a unique combination of
phrases.
- No new phrase for PNO 3 (applying 2 above.)
since PNO 3 appears in the same combination as
PNO 2.
PNO 1 PEO 7 PEO 8 PEO 10 PNO 2 PNO 4 PEO 4
PEO 9 PEO 6
PNO 3 PNO 5
Figure 7.A-54
New Phrase Merged from Phrases 1, 2, and 3
When a value for the merge key appears in only one phrase,
that phrase is copied into the new record. The phrase
is inserted for every different value that occurs uniquely
in that phrase.
Thus, phrases 3, 4 and 5 are inserted in the new
record as individual phrases since they contain PNO's
4, 6, and 7 respectively. Notice that Phrase 3 is not
inserted redundantly for the occurrence of PNO 5.
All other phrases in the record are inserted in the
new record.
Multiple merge keys may be specified as DATA
entries. Phrases within a record are merged into a new
phrase if, and only if, they contain the same values
for the combination of merge keys.
USER LANGUAGE MANUAL
Inter-record & Inferential Logic
7.A.5.2.2.
-277-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Figure 7.A-55
Ne:;4 Raz..02d after MERGEP Execution
ORD
ONTROL
HEAMER
Merged
Ph. 1
PNO 1
PEO
rmi 8
Ja0 IL
PNO 2
PNO 4
Merged
Ph. 2
PNO 1
PEO 7
PEO 8
PNO 2
PEO 4
PEO 9
Ph. 3
PNO 1
PEO 10
PNO
PNO 6
PNO 5
Ph. I.
PNO 6
PEO 1
PNO
PEO
PNO -
Ph. 5
PNO 7
PEO 14
PNO
PEO
PEO -
Ph. 6
PNO -
PEO
PNO
PEO -
Ph. 7
PNO
PEO
PNO
PNO
PEO -
-1
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
1 1 1 ' 1 1 1 1 1 1 1 1 1
rn
r)
A,
rn
-4
Approved For Release 2000/05/NCRCIA-RDP78-03952A000100070001-5
Phrases 1 and 2 in Figure 7.A-56 contain common
values for the merge keys PNO and PEO, namely, PNO 1 and
PEO 7. Phrases 1 and 2 will therefore be merged into a
single new phrase. Phrases 2 and 3 contain common
value 3 for merge key PNO, but since the values for the
second merge key, PEO, differ, they are not merged.
All other points above apply except that a common
value for both merge keys must be present.
7.A.6. FILE MAINTENANCE
Two types of file maintenance transactions are
distinguished in the CHIVE System:
- Conditional file maintenance--changes to CHIVE
files involving data elements which apply to
logically specifiable classes of data elements.
(Specification of these changes involves the use
of the CHIVE language in a way which is similar
to its use for retrieval transactions.)
- Unconditional file maintenance--changes to CHIVE
data elements which can be specified through the
use of unique control terms: record control,
phrase control, tag name, or tag name-value.
7.A.6.1. Conditional Maintenance
Conditional maintenance is accomplished through the
use of APPEND, DELETE, and REPLACE commands.
The Rules and Explanations for each of these commands
are identical to those given in Section 7.8. for the
same commands of the query language, i.e., for retrieval
transactions.
USER LANGUAGE MANUAL
File Maintenance
-279- 7.A.6.1.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 200011137118T: CIA-RDP78-03952A000100070001-5
Figure 7.A-56
MERGE Phrases for Multiple Merge Keys
RECORD
CONTROL
HEADER PHRASE
PNO 1 PEO 7 PNO 2
PNO 1 PEO 7 PNO 3
PNO 2 PEO 2 PNO 3
PNO 3 PEO 4 PNO 5
PNO- PEO- PNO -
PNO- PEO- PNO-
-
Ph.
Ph. I
Ph. 2
Ph. 3
Ph. 4
Ph. 5
Ph. 6
-280-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0SECM-RDP78-03952A000100070001-5
7.A.6.1.1. APPEND Command
This command adds the data which appears as a DATA
entry to the data element that meets the search CRITERIA.
COMMAND LIMIT CRITERIA DATA
R P T
1 APPEND
2 APPEND
Z C PNO 7641
A PNO 7294 & PEO 7396
Figure 7.A-57
Command Lines for File Maintenance APPEND
Rules:
1. Search CRITERIA for records (R) and phrases (P)
are required to APPEND terms. To APPEND phrases to
records, only record (R) CRITERIA entries are necessary.
2. File input and output entries are not required.
The file being maintained is named in the MAINTENANCE JOB
SPECIFICATIONS (see Section 7.A.3.).
3. All data to be appended must appear as DATA entries
4. The LIMIT entry is optional.
5. Records cannot be appended to the file with this
command.
Explanation
Figure 7.A-57 shows two examples of APPEND commands:
(1) APPEND adds "PNO 7641" to phrases which meet the
phrase CRITERIA entry, i.e., the condition specified on
USER LANGUAGE MANUAL
File Maintenance
Approved For Release 2000/05/MC1A-RDP78-03952403:1011=70001-5
SECRET
Approved For Release 20003135REr : CIA-RDP78-03952A000100070001-5
test line C, in all records. (2) APPEND adds the phrase
"PNO 7294 PEO 7396" to all records which meet the record?
CRITERIA entry A.
7.A.6.1.2. DELETE Command
Terms, records, or phrases are deleted from the file.
COMMAND LIMIT CRITERIA FILES DATA
RP T
I DELETE
2 DELETE
3 DELETE
A Z C
Figure 7.A-58
Command Lines for File Maintenance DELETE
Rules:
1. CRITERIA entries are required as follows:
(R) - to DELETE records specified
(R) (P)
- to DELETE specified phrases from
specified records.
(R),(P),(T) - To DELETE specified terms from
specified phrases in specified
records.
2. No DATA entries appear.
3. Other entires are the same as the APPEND command.
Explanation
Figure 7.A-58 shows examples of three DELETE commands:
(1) DELETE - terms which meet conditions on test line C
-282-
USER LANGUAGE MANUAL
Approved For Release 2000/05/08 : CIA-RDP*-ktplqAptffitieg60001-5
SECRET ? ? ? ? ?
Approved For Release 2000/0511KMIA-RDP78-03952A000100070001-5
are deleted in all phrases of records which meet condition
specified on test line A. (2) DELETE - phrases specified
on test line B are deleted from all records. (3) DELETE -
records which meet conditions specified on test line D
are deleted from the file.
7.A.6.1.3. REPLACE Command
Data elements in the DATA entry REPLACE those
data elements which meet the search CRITERIA.
COMMAND LIMIT CRITERIA FILE DATA
R P T
1 REPLACE
2 REPLACE
B C PEO 17
A B PNO 2617 & PEO 99
Figure 7.A-59
Command Lines for File Maintenance REPLACE
Rules:
1. Search CRITERIA entries are required as follows
(R).(P) - to replace a phrase
(R),(P),(T) - to replace a term
2. Records cannot be replaced.
3. Other rules are identical to the rules for the
APPEND command.
USER LANGUAGE MANUAL
File Maintenance
7.A.6.1.3.
-283-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRe
Approved For Release 2000386W: CIA-RDP78-03952A000100070001-5
Explanation
Figure 7.A-59 shows examples of two REPLACE commands:
(1) REPLACE - the term in the DATA entry replaces all
terms which meet term, phrase, and record CRITERIA
entries. (2) REPLACE - the phrase in the DATA entry
replaces all phrases which meet phrase and record CRITERIA
entries.
7.A.6.2. Unconditional Maintenance
A special form will be provided to specify maintenance
to be performed when a uniquely specifiable element of a
file is addressed. This form will include (among other
things) on each line:
- the unique record to be addressed (by record
control terms)
- the unique phrase to be addressed (by phrase
control data)
- the unique term to be addressed (either the tag
name or a tag-value combination)
- the operation to be performed
replace)
(append, delete,
- the data to be inserted or appended.
USER LANGUAGE MANUAL
File Maintenance
7.A.6.2.
-284-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/085RDP78-03952A000100070001-5
Appendix 7.B.
INPUT TRANSCRIPTION STUDY
One of the major problems of the CHIVE system is that
of reading the indexing records and other data into the
computer system for subsequent processing and file
maintenance operations. Four categories of systems were
evaluated for this application: (a) punched cards,
(b) punched paper tape, (c) cathode-ray tube/keyboard
(CRT/KB) equipments, and (d) page readers. The results
of that evaluation are described in the sections that
follow, using the above headings.
A preliminary oral presentation of the major portion
of the material included in this report was made to OCS
and CHIVE staff members on 19 November 1964. Immediately
following that presentation, a decision was made jointly
by the OCS and the CHIVE staffs to utilize the page
reader methods for the indexing operation, and to give
further attention to that specific area. Accordingly,
supplementary investigations were made (primarily by
phone) to provide a broader base of information in
anticipation of a page reader procurement program.
INPUT TRANSCRTPTION
7?B.
-285-
Approved For Release 2000/05/08 .? CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/WFM-RDP78-03952A000100070001-5
The fundamental assumptions on which the evaluations
were based are specified in Section 7.B.1, and details
concerning operational aspects and the cost and manpower
requirements of each of the candidate methods are
included in Section 7.B.2. The results of the study are
summarized in Section
costs for each of the
in CHIVE/R-2-65.
7.B.3.,
systems
and a breakdown of the
configurations is provided
7.13.1. FUNDAMENTAL ASSUMPTIONS
The various candidate input methods differ somewhat
in detail, but there are certain basic procedures and
factors that would be common to all of the methods.
The following assumptions were made for all evaluations:
- The index records are provided in a format
(indexer work sheet) that is to be used for direct
transcription into the data-input system.
- The index records are transcribed with keyboard
equipment (where proofing is required, hard copy
is also prepared during this operation).
- Proofing is made with reference to the original
indexer work sheet.
- Corrections are made as necessary against the
proofing information.
- The index records are transcribed off-line onto
magnetic tape for subsequent computer processing.
INPUT TRANSCRIPTION
Fundamental Assumptions
7.B.1.
-286-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/04ctiR-RDP78-03952A000100070001-5
- All equipment operating speeds are de-rated to 75%
of the corresponding nominal values, unless
estimates of effective rates were available from
other sources.
- Equipment costs were determined both for purchase
(including maintenance), based on five-year
amortization at zero percent interest rate and
for lease. The lower of the two costs is used
for the evaluations in each case.
Cost information is provided graphically in
Section 7.B.2. in terms of the program cost per year,
based on a variable number of index items processed.
Salaries for experienced typists are assumed to be $5170
per year, salaries for keypunch operators $4650 per year,
and salaries for special equipment operators $6000 per
year for a single-shift operation of 264 days per year.*
Costs for quantitity items such as cards, paper, etc.,
are included, but in no case are overhead costs, or costs
that are believed to be common to all candidates
incorporated into the estimates.**
The cost and the manpower requirements for the
transcription of the ,index records are very significant.
Accordingly, the effective production rates are
correspondingly critical. Based on ihforma ion obtained
* C.F., CHIVE/R-2-64
** A cost breakdown for each system is provided in
CHIVE/R-2-65.
INPUT TRANSCRIPTION
Fundamental Assumptions
7.8.1.
-287-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0eqlg-RDP78-03952A000100070001-5
from various sources,* the production rates assumed for
the evaluations were as follows:**
Typewriter--10,000 characters per hour
Paper-tape punches and IBM 826--9000 characters per hour
Keypunch--8000 characters per hour.
In all evaluations, it was assumed that there are 1000
characters per indexing record, and that only upper case
characters are utilized.
On visits made to the Defense Documentation Center (DDC)
and to the National Library of Medicine (NLM) immediately
* The IBM Typewriter Division specifies a typical
typewriting rate of 10,200 characters/hr; personnel
implementing a new library indexing operation at
Lockheed, Palo Alto have determined that production
rates slightly in excess of 10,000 characters/hr. are
achieved by operators working for periods of two hours
on the IBM 826 keyboard; in a report entitled "Application
of Print Readers to the Needs of Intelligence Agencies,"
Study-147, March 1964, the Institute for Defense Analysis
reports that an experienced keypunch operator can punch
about 500 verified cards/day. If it is assumed that there
are 80 characters punched per card, and that the verifier
rate is 40% faster than the keypunch rate, the production
of 500 verified cards/day would correspond to a keypunch
rate of approximately 8000 characters/hour. Continuing
OCR/MD tests have also indicated keypunch rates of
approximately 8000 characters/hour.
**In all evaluations, costs were calculated and are
presented graphically for typewriter/keypunch speeds
varying over a range of approximately + 12% from these
nominal values.
INPUT TRANSCRIPTION
Fundamental Assumptions
7.B.1.
-288-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/gcla-RDP78-03952A000100070001-5
prior to the oral presentation at the Agency, information
was obtained on production k rates achieved at those
organizations for operations where the keyboard operator
is not provided complete information from a single work sheet,
but is instead required to extract material to be transcribed
from both the work sheet and the original document or journal.
For these cases, the rates were reported to be approximately
only 3500 characters per hour. Certainly for direct
transcription from a single work sheet, the average rate
achieved must be much higher than this figure, but the
problem of predicting a reasonable transcription rate
should be considered further. Better estimates could be
obtained with an in-house test run for several days
with A representative group of typists and keyboard
operators working on representative input records.
The, proofing operation is reported to constitute a
very significant time requirement at Lockheed, at DDC, and
at INLM where the hard copy produced during the transcription
process is checked directly against the original. Based
on the Lockheed figures, it was specified for our
eva-luations that the proofing operation requires 2.8 minutes
INPUT TRANSCRIPTION
Fundamental Assumptions
7.B.1.
-289-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0?KW-RDP78-03952A000100070001-5
per indexing record.* The (normalized) times required
for proofing at DDC and at NLM were reported to be
greater than that required at Lockheed by approximately
the same factor as the ratios of the times required for
the corresponding transcription operations (approximately
2.6:1).
The greatest variable of all the operations in the
various inputting systems is that of the correction time
required. For example, in the paper tape and the
punched card cases where normal correction procedures
involve scanning the original tape or cards to locate the
specific items that must be corrected, the time requirement
is severe. Conversly, with the cathode-ray tube/keyboard
(CRT/KB) technique, the text is immediately available
on the display tube for processing, and the time required
to achieve the necessary corrections is minimized. For
the evaluations, the correction time requirement was
assumed to vary over the range from 2.8 to 0.8 minutes
per indexing record. The correction times are specified
individually for each input system considered in the
* This specification applies to all systems except the
punched card case where the IBM 059 Verifier is used.
For this case, initial OCR/MD test data is used.
INPUT TRANSCRIPTION
Fundamental Assumptions
7.B.1.
-290-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/gcb1A-RDP78-03952A000100070001-5
following sections of this report.* It was assumed
here that there are two errors per item. This assumption
was based on the statement by the IBM Typewriter Division
that an experienced typists makes one error in approximately
500 characters, and on the statement by the DDC personnel
that their typists typically make two errors in
approximately 80 percent to 85 percent 6f the 1000-character
items transcribed (NLM specifies that errors occur in only
27% of the items produced by their Flexowriter operators,
but these items are
of 413pr9t,ply pr4y FO characters
in length. Complete statistical data are not available
from NLM).
As noted previously, the decision vas made after
the preliminary oral presentation to plan on the use
of page readers. Coupled with this decision are new
operational aspeCtS of the input problem that axe to be
considered for a slightly modified mode of operation:
-The output of the page reader is to be coupled
directly to the master data processing system by
the executive control system, rather than being
transferred off-line to magnetic tape. This
eliminates the need for a separate magnetic tape
oar system associated with the page reader, but adds
the need for some interface equipment.
*The primary reference data utilized is that provided
by the Lockheed staff in charge of implementing the
Library indexing and transcription operation.
INPUT TRANSCRIPTION
Fundamental Assumptions
7.B.1.
-291-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08S:E6MIRDP78-03952A000100070001-5
- Whereas the original page-reader evaluations were
based on single-font equipments, any equipment to
be considered for future evaluations should have
the capability of reading two or more fonts.
- In those cases where an indexer may elect to use a
typewriter in the preparation of the index work
sheets, it should be possible that these work
sheets may be used directly as the input document
to the page reader system, thereby eliminating
the need for a separate typing operation. It is even
possible that for the long term, with a suitable
education program, all of the indexers can be induced
to prepare their work sheets with a typewriter in
a form suitable for direct scanning by the page
reader. The additional cost of the extra time that
would probably be required by the indexers for the
preparation of suitable typewritten work sheets would
necessarily be compared to the cost of the typewriter
transcription process that has thus far been assumed
to be necessary. However, it would seem to be an
area in which some additional cost and manpower
savings might be realized.
In Section 7.B.2.4., in which the page reader operation
is considered, the costs and the manpower requirements are
specified primarily for the original mode of operation.
However, cost data are provided also for the on-line mode
of operation with the one currently-available, production
page reader capable of operation on a multi-font basis.
7.B.2. EVALUATIONS OF THE CANDIDATE INPUT SYSTEMS
7.B.2.1. Punched Card Systems
Punched card systems of all shapes and sizes have been
in operation essentially since the beginning of automatic
INPUT TRANSCRIPTION
Evaluations
7.B.2.1.
-292-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08SWRDP78-03952A000100070001-5
data-processing systems. Accordingly, a detailed discussion
of the punched card systems as possibly applied to the CHIVE
data input operation is not warranted in this report. It
will suffice to only review briefly the three configurations
of punched card systems that were considered in the project
studies. These three systeMs are:
- A keypunch/verifier configuration. It is to be
noted that the costs considered in this evaluation
are those of the recently announced solid-state
models of the keypunch and verifier (IBM 029/059
equipments) instead of the familiar IBM 026/056
equipments. This caused only a slight percentage
increase in the overall cost of the system.
- A keypunch/listing configuration. In this case
a keypunch operator punches the cards in the normal
fashion, but uses a listing for a visual proofing.
- An IBM 826 configuration. In this case hard copy
is prepared on a typewriter electrically coupled
to a companion keypunch unit, on which cards are
simultaneously punched. (This is the configuration
utilized very successfully by the Lockheed staff
responsible for implementing a new document
cataloging and retrieval system in their library.)
The summary cost information for the three punched card
systems is shown in Fig. 7.B-l. Operation of the keypunch/
verifier configuratiQn is the most expensive of the
three--primarily because more opetators and more pieces
of equipment are reqUired (c.f., the comparative tabulation
ill table 7.B-2, Seton 7.8.3., in which the number of
typists or key pundh operators required is listed). The
INPUT TRANSCRIPTION
Evaluations
703.2.1.
-293-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
ANNUAL COST ($)
Approved For Release 2000/05WFM-RDP78-03952A000100070001-5
IBM 029/059
IBM 029/ LISTING &
IBM 826
500,000 1,000,000
INDEX ITEMS/YEAR
OPERATING COSTS
PUNCHED CARD SYSTEMS
FIGURE 711-1
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/01-RDP78-03952A000100070001-5
other two configurations are approximately equivalent in
terms of operating cost, as indicated in Figure 7.B-1.
As stated in Section 7.8.1., the proof time that
was utilized in all of the cost evaluations is 2.8
minutes per indexing record, except for the keypunch/
verifier equipment configuration. In this case, the
proof was specified to be 6.4 minutes per record, this
being based on initial estimates of keypunch verifying
rates.
Regarding the correction operation the keypunch/
listing operation was specified to require a relatively
large
item.
amount of time for
This was based on
for the relatively large
corrections-2.6 minutes per
estimates of the time required
number of steps involved. In
particular, since efficient listing operations require
that the cards from many items be run through the lister
in large batches, a considerable amount of time is
subsequently required for scanning for, and retrieval of,
the specific cards for which Correction cards must be
made. For the keypUnoh/verifier (029/059) and the IBM
826 operation, procedures can be arranged whereby the cards
for a specific item can be retained by the operator(s)
until corrections are made. Accordingly, the time
INPUT TRANSCRIPTION
Evaluations
7.B.2.1.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05KWRA-RDP78-03952A000100070001-5
required for corrections can be minimized to a reasonable
extent. For these two equipment configurations, it was
specified for the evaluations that a correction time of
1.6 minutes per item is required.
For the listing operation and for the conversion
from punched cards to magnetic tape, two modes of
operation were assumed. For relatively low rates of
production, rates quoted by service bureaus were utilized
in determining costs. However, for relatively high
production rates where cost analysis indicated justification
for purchase of suitable equipment, costs were calculated
on that basis. In the latter case, purchase of a Univac
Model 1004-2 was assumed for listing operations, and
purchase of CDC equipment was assumed for the card-to-
magnetic-tape conversion (c.f., CHIVE/R-2-65). In both
cases, the cost of operators for special machines was
included.
The number of keypunch operators (or typists) and
proofing clerks required for the three punched card
systems for production of from 100,000 to 1,000,000 items
per year varies from 8 to 129. A complete comparative
tabulation is provided in Section 7.B.3. (Table 7.B-2).
INPUT TRANSCRIPTION
Evaluations
7.B.2.1.
-296-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/6PWA-RDP78-03952A000100070001-5
7.B.2.2. Punched Paper Tape Systems
Punched paper tape equipments have been developed to
a relatively high degree of sophistication and reliability
during the last few years. In some cases, manufacturers
have incorporated IBM Selectric typewriters as the basic
element of their equipments to increase the operating
ease and efficiency. From the automatic control standpoint
various manufacturers have incorporated program control
features into their equipments whereby special paper
tape loops punched with specific program Control codes
can be used to greatly facilitate repetitive modes of
operation. Six types of units were investigated in this
study.
The result of the cost evaluations for the punched
paper systems are shoWn in Figs. 7.8-2 and 3. Por all such
equipments considered, the cost variations are relatively
*The equipments investigated were the Dura Mach 10,
Priden Plexowriter, the Greene Letteriter, the Royal
SDW 5000, the Teletype Model 33, and the Teletype Model 35.
Additional units such as the Xleinschmidt Model 321, and
the RemingtOn Rand Synchrotape units are available, but
were not evaluated. In the case of the Synchrotape,
Remington Rand has sold All inventory and production
rights to Radio Electronics, Inc., New York City, and
this latter company is withholding further production
until sales trends of their new magnetic tape typewriter
units are evaluated.
INPUT TRANSCRIPTION
Evaluations .
7.B.2.2.
-297-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/085%1MIRDP78-03952A000100070001-5
1,000,000
a-
???????
cr)
0
(.)
-J 500,000
c:t
111111111
GREENE, DURA MACH 10,
a FRIDEN
...............
I I
TELETYPE No. 33
0 500,000
INDEX ITEMS/YEAR
OPERATING COSTS
PAPER TAPE SYSTEMS
FIGURE 7.6-2
1,000,000
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/054696TA-RDP78-03952A000100070001-5
7.8.2.2. Punched Paper Tape Systems
Punched paper tape equipments have been developed to
a relatively high degree of sophistication and reliability
during the last few years. In some cases, manufacturers
have incorporated IBM Selectric typewriters as the basic
element of their equipments to increase the operating
ease and efficiency. From the automatic control standpoint
various manufacturers have incorporated program control
features into their equipments whereby special paper
tape loops punched with specific program Control codes
can be used to greatly facilitate repetitive modes of
operation. Six types of units were investigated in this
study.*
The results of the cost evaluations for the punched
paper systems are shown in Figs. 7.8-2 and 3. ?or all such
equipments considered, the cost variations are relatively
*The equipments investigated were the Dura Mach 10,
Friden Flexowriter, the Greene Letteriter, the Royal
SDW 5000, the Teletype Model 33, and the Teletype Model 35.
Additional units such as the Kleinschmidt Model 321, and
the Remington Rand Synchrotape units are available, but
were not evaluated. In the case of the Synchrotape,
Remington Rand has sold all inventory and production
rights to Radio Electronics, Inc., New York City, and
this latter company is withholding further production
until sales trends of their new magnetic tape typewriter
units are evaluated.
INPUT TRANSCRIPTION
Evaluations
7.8.2.2.
-297-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
ANNUAL COST ($)
Approved For Release 2000/05/08E0M-RDP78-03952A000100070001-5
1,000,000
500,000
F:Kel
GREENE, DURA MACH 10,
a FRIDEN
TELETYPE No. 33
I I
500,000
INDEX ITEMS/ YEAR
OPERATING COSTS
PAPER TAPE SYSTEMS
FIGURE 7.6-2
1,000,000
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
ANNUAL COST ($)
1,000,000
500,000
ROYAL
TELETYPE No. 35
I I I I I I I I
500,000
INDEX ITEMS /YEAR
OPERATING COSTS
PAPER TAPE SYSTEMS
FIGURE 7.B-3
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
1,000,000
Approved For Release 2000/05/08EGRIA-RDP78-03952A000100070001-5
small--approximately 10 to 15 percent. This is due primarily
to the fact that the major costs are the salaries of the
typists, thereby subordinating the effects of the variable
equipment costs, even though the various units range in
purchase price from $652 to $4,000.
Large-scale data-entry installations utilizing
punched paper tape equipments such as those visited at
the Defense Documentation Center (DDC) and the National
Library of Medicine (NLM) have been in operation for a
significant period of time, and satisfactory operational
results are reported. However, it appears that the use
of punched paper tape systems is declining rather than
expanding. The administrative personnel in charge of
several such installations stated that they are considering
a change-over to some data-input system other than the
punched paper tape system because they believe that
improvements in the overall efficiency of operation can
be achieved. In addition, discussions with the data-
processing personnel at a large title insurance company
produced the information that they consider punched paper
tape to be operationally unsuitable. In their opinion,
the problem of making corrections is a difficult one,
and in addition, merely the problem of handling very
INPUT TRANSCRIPTION
Evaluations
7.B.2.2.
-300-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05Acta-RDP78-03952A000100070001-5
large amounts of paper tape in either fan-fold or roll
form is too cumbersome to be considered feasible.
Accordingly, they have decided to utilize page readers
rather than punched paper tape equipments.
The correction times specified for use in these
evaluations of the punched paper tape systems is 2.8
minutes per index item. It is to be noted that this is
the largest time requirement specified for any of the
systems considered. ThiS specificatlion was based
primarily on the problems of achieving corrections of
errors that are made at random points throughout the
material. In the gttaightforward correction method such
as that used at DDC, it is necessary to utilize the
initial tape f?r automatic re-typing and re-pundhing Of
the item to the po4it where the first Correction Must
made--at this point the operator stops the automatic
operation of the equipment, and types in the correction
material. The operator then again places the machine
into the automatic operation mode until the next error
?
is encountered, or until re-typing and re-punching of
the item is completed. Perhaps some improvemant can be
achieved by utilizing a more sophisticated semi-automatic
correction technique such as that employed at NLM, but
INPUT TRANSCRIPTION
Evaluations
7.8,2.2.
-301-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/088131M-RDP78-03952A000100070001-5
unfortunately, adequate information concerning time
requirements for the techniques utilized there are not
now available. The procedure at NLM is to prepare a
special correction tape rather than using the straight-
forward method of preparing a new, corrected tape of the
index item. The correction tapes must contain full
identification of the item, the specific designation of
the location within the index item where the correction(s)
must be made, and the designation of the nature and
content of the correction(s) to be made. At NLM, the
correction tapes are then spliced into the original tapes
containing the index items, and all information is then
read into the data-processing system at one time. The
actual corrections are then made automatically by the
computer. The computer is also programmed to check the
text material for specific types of errors such as those
which might have been made in specification of descriptors
etc., and a printout of these additional errors is made
for use in preparation of a second set of correction
tapes. This additional correction information is then
punched into tape, and entered into and processed by the
computer, but no final check of the data entered is made.
INPUT TRANSCRIPTION
Evaluations
7.B.2.2.
-302-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0PCN-RDP78-03952A000100070001-5
For the conversion from paper tape to magnetic tape,*
the use of a service bureau at standard rates was assumed
for relatively low rates of production. However, fOT
higher production rates where the service bureau costs
could prove to be excessive, purchase of the necessary
conversion equipment was assumed (c.f., CHIVE/R-2-65).
The cost of an operator for the conversion equipment was
also
from
incorporated into the cost analysis.
For the transcription, proofing, and correction
100,000 to 1000,000 index sheets per year, the
number of typists and proofing clerks required ranges
from 9 to 103. A complete tabulation is provided in
Section 7.B.3. (Table 7.B-2).
?of
7.B.2.3. The Cathode-Ray Tube/Keyboard (CRT/KB) System,
A greatly expanded utilization of the cathode-ray
tube has been made for information display and manipulation
during the past few years. Of the earlier special
*Both IBM and Radio Electronics, Inc., have recently
introduced keyboard equipments for data storage directly
on magnetic tapes. However, these units Were not
considered extensively in this study. Neither unit
produces a tape that can be utilized on computer tape
transports, and it is believed that the operational
limitations and disadvantages of punched paper tape
systems would be common to these new magnetic tape
systems.
INPUT TRANSCRIPTION
Evaluations
7..2.3.
-303-
Approved For Release 2000/05/08 .? CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05fac:REA-RDP78-03952A000100070001-5
applications of cathode-ray tubes, perhaps one of the
most notable was the SAGE system where the cathode-ray
tube is used as the display medium in conjunction with
radar tracking of aircraft. This system incorporated
character and symbol generation, and light pen techniques
experimentally as early as 1954. Since that time,
cathode-ray displays have been used in relatively
sophisticated form in conjunction with computers of
various manufacturers to expedite entry and analysis of
numeric or alphabetical data stored in memory. It is
only natural that techniques and equipments of this
general nature should ultimately find widespread
application in the general fields of data entry into,
and data retrieval from, data-processing systems.*
Equipments produced by six corporations have been
*There are at least six such CRT/KB systems reported to be
installed for commercial use at this date. Also, smaller
desk sets have been marketed for specialized uses such as
stock price and activity quotations.
INPUT TRANSCRIPTION
Evaluations
7.B.2.3.
-304-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/8FitTA-RDP78-03952A000100070001-5
evaluated in connection with the study of the CHIVE data-
input problem.*
The basic configuration of the CRT/KB system is very
simple. Fundamentally, it consists merely of a typewriter-
like keyboard with which the operator enters data into the
system. As he types, the characters are "instantaneously"
displayed on the face of the oscilloscope, and are
simultaneously stored in a temporary buffer register.
Generally, some form of electronic pointer or cursor is
proVided to designate on the face of the display tube the
position at which the next character shall be displayed or
revised, Ideally, the central data-processing unit is not
required in connection with any of this initial data-entry
process associated with the CRT/KB unit. It is only when
the operator has completed typing all of the data involved,
and has checked and corrected it as required, that the
central processing unit is involved in any way. At this
*These dorporations are Burroughs, Data Display Inc., IBM,
Information Products Corporation, Philco, and Teleregister.
However, it should be noted that the Burroughs and Philco
equipments ard not production items, and that the corresponding
pricing information is not firm. Other companies are also
producing equipments capable of character generation and
oscilloscopic display. Raytheon, Inc. has recently
announced a very sophisticated system designed primarily for
military applications. Raytheon plans to produce a less
expensive version for commercial applications, but is not
yet prepared to provide any technical specifications, or
cost and delivery schedules.
XNPUT TRANSCRIPTION
Evaluations
-305-
Approved For Release 2000/05/08: CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0gfflfgRDP78-03952A000100070001-5
time, the operator presses some form of a data-entry key,
thereby indicating to the system that, as appropriate, the
central data-processing unit should be interrupted for
acceptance or processing of the data stored in the CRT/KB
buffer.
One of the major features of the CRT/KB system is
the facility with which corrections can be made. In all
of the equipments considered in this study, simple
"substitution" types of corrections can be made merely
by moving the display pointer until it designates the first
character to be corrected, and then typing the correct
characters so that the operator can check them directly,
and the corresponding changes are simultaneously made in
the buffer storage. An even greater facility is realized
in those cases where insertion of characters, words, or
even sentences must be achieved at some point within the
body of text that has already been entered into the
CRT/KB system. In this case, the operator needs only to
designate the point of insertion, press some type of
"Insert" key, and then type the characters that are to be
entered. The text material following that which is being
inserted is automatically advanced--both on the face of
the display tube and in the buffer storage. This
facility is incorporated within the CRT/KB units produced
INPUT TRANSCRIPTION
Evaluations
7.B.2.3.
-306-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08EW-RDP78-03952A000100070001-5
by some of the manufacturers,* but in other cases it
would necessarily be provided in conjunction with access
to and special programming for the central data-processing
unit.
The cost information for the CRT/KB units is
provided in Figures 7.B-4 and 5. It is to be noted that
the cost for a system based on the IBM 2250 equipments
is very much greater than that of any of the other units.
This IBM equipment is much more sophisticated than is
required for the CHIVE input problem, but it was considered
mei desirable to include it, nevertheless, for comparative
purposes. The unit is capable of as many as 3,848
character positions**onithe face of the tube, and provides
am the capability of graphical display and light-pen operations.***
*The insertion capability is provided in the Burroughs,
IBM and Philco equipments.
MO
owl
**However, where as many as eight display stations are
connected to one control unit, the number of characters
per station which can be simultaneously displayed may
be limited to as few as approximately 1,000 characters.
IBM has not yet made information of this type generally
available.
***IBM also provides the Model 1015 type of unit, but this
unit is designed primarily for use in data inquiry
applications. As many as 60 Model 1015 inquiry stations
can be operated in conjunction with on control unit (Type
1016), but only one station can be active at any one time.
Consequently, it would not.be suitable for a data-entry
application of the type considered in the study. Also,
the display tube is of the storage type; this would
probably introduce some minor complications in connection
with correction of errors.
INPUT TRANSCRIPTION
Evaluations
7.B.2.3.
-307-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0?NWRDP78-03952A000100070001-5
1,000,000
500,000
IBM 2250
BURROUGHS
TELEREGISTER 203 a
IPC 1560
500,000
INDEX ITEMS/YEAR
OPERATING COSTS
CRT/KB SYSTEMS
FIGURE 7.B-4
1,000,000
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
MP
_J
4
mei
esti
?
gni
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
1,000,000
500,000
PHILCO
dd 10
500,000
INDEX ITEMS /YEAR
OPERATING COSTS
CRT/KB SYSTEMS
FIGURE 7.B-5
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
1,000,000
Approved For Release 2000/05/08 tagRDP78-03952A000100070001-5
Of the units manufactured by the other corporations, the
costs of processing one million index items per year do
not differ by a significant amount. However, for processing
of 100,000 items per year, the cost of the Burroughs and
the Philco equipment is substantially greater. This is
because of the fact that those system designs are based
on large central control units which are required regardlOss
of the number of data entry and display stations connected
into the system--as contrasted to designs on a modular
control-unit basis.
? For the entry, proofing, and correction of from
100,000 to 1,000,000 index sheets per year, the number
of typists required ranges from 7 to 8. A complete
tabulation is provided in Section 7.B.3. "(Thble 7.B-2).
This constitutes the lowest typist requirement? of all
of the systems considered, because of two factors.
First, typists (as compared to slower keypunch operators)
are utilized for the entry of the data into the system.
Secondly, evaluations were based on the specification
that the correction time required by the CRT/KB systems
is shorter than it is for any of the other systems
considered. In this case, a correction time of 0.8
minutes per indexing record was specified.
INPUT TRANSCRIPTION
Evaluations
7.8.2.3.
-310-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
noi.4
Approved For Release 2000/05/0F6U-RDP78-03952A000100070001-5
7.B.2.4. Page Reader Systems
Commercially produced optical character readers have
been available for approximately ten years. However, it
has been only in the last two or three years that such
readers have become relatively sophisticated, configured
to read full pages of printed information (as well as
single lines in rigidly controlled locations on the
document), and capable of providing a reading accuracy
equal to or better than one reject per thousand characters
and one substitution per 10,000 characters (with relatively
high-quality typewritten material).* Improvements in
accuracy have been due not only to improved techniques of
character recognition, but also, to a significant extent,
to the cooperation of typewriter manufacturers in
exericising better control of the characters produced by
their typewriters. Previously, the control of the
alignment and the detail of the characters had been
inadequate.
During the past few months, the character reading
mid situation has reportedly been improved even further.
*Higher accuracies are generally obtained with characters
produced with high-quality printing equipments. Printing
of this type is termed "pre-printed" material, and type-
written characters fall into the "post-printed" category.
INPUT TRANSCRIPTION
Evaluations
7.B.2.4.
-311-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/085%MIRDP78-03952A000100070001-5
Unfortunately, these reports haves not yet been backed up
by publications, but they were provided by reliable
sources and are therefore considered to be of significance.
In brief, it was stated that during recent acceptance tests
of a commercial page reader, there were no substitutions
whatsoever, and the reject rate was so low that tabulation
of rejects was discontinued in deference to the operational
aspects of the tests. These tests were made with relatively
high-quality print control, but it seems reasonable to
infer that reading of printing of somewhat lower quality
should be possible at a moderate sacrifice of accuracy.
At the present time, only two companies are offering
page readers for sale that can be considered to be
production equipments. These companies are Farrington
Electronics and Recognition Equipment. Farrington has
delivered at least ten page-reading units capable of
reading a single font--the Self-Check font--and Recognition
Equipment has delivered four page-reading units capable of
reading such fonts as IBM 407E, IBM 1403, and E13B. In
addition, Farrington has delivered approximately 90
document readers, and Recognition Equipment has delivered
at least one.
INPUT TRANSCRIPTION
Evaluations
7.B.2.4.
-312-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
NI
Approved For Release 2000/05/CfEECOA-RDP78-03952A000100070001-5
Two other companies--Rabinow Electronics and Philco
Corporation--are scheduling deliveries of their new types
of page readers during the first or second quarter of 1965.
The Rabinow equipment is to be capable of reading only the
ASA font, but the company personnel claim the capability
of producing a multi-font machine on special order. The
Philco machine is to be capable of reading several fonts--
the maximum number being of the order of approximately 16.
However, it must be noted that this machine is not really
a multi-font machine capable of reading intermixed fonts
without some sort of control being exercised. Specifically,
each field of text printed with characters of a single type
of font must be designated by means of plus signs, or the
machine must be programmed to shift the recognition
programs for arbitrarily specified, pre-designated fields
of text, there being only one type of font per field.*
In addition to the current or imminent producers of
page readers as enumerated above, GE, IBM, and RCA appear
to have plans for marketing commercial page readers in
the near future. However, no significant amount of
*There is one exception--more than one font can be read
simultaneously in a given field provided that corresponding
characters of all such fonts have identical aspect ratios.
INPUT TRANSCRIPTION
Evaluations
7.B.2.4.
-313-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/088ECRURDP78-03952A000100070001-5
information concerning machine characteristics, and costs
and delivery schedules, is available for evaluation (as of
December, 1964). In a final category are Sylvania, National
Cash Register, Burroughs, and Link--all of which have
character reader programs under way, but no announced or
implied intention of producing a "standard" type of page
reader within the foreseeable future.*
It is not appropriate in this report to attempt to
provide detailed information concerning the basic features
and limitations of each of the candidate page readers.**
Investigations and evaluations in depth would be more
appropriate during a procurement phase of the CHIVE
program. Further considerations are limited here to the
cost and manpower aspects of the anticipated CHIVE indexing
program.
Cost information for single-font machines is
provided for off-line page-reader systems incorporating
the Farrington reader, the Rabinow reader, and the
* However, the Sylvania personnel have indicated that they
would be glad to receive any request for bid that may be
issued in conjunction with procurement of a page reader.
**The SRI project personnel have on file brochures and
contact reports concerning the salient details of the
page readers that have been developed to a reasonable
degree at this time.
INPUT TRANSCRIPTION
Evaluations
7.B.2.4.
-314-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0tEW-RDID78-03952A000100070001-5
Recognition Equipment reader in Figures 7.8-6, 7, and 8,
respectively. In each case, cost information is
provided for two sets of conditions. The upper band
is required to pay the full cost of the page-reading
equipment. The lower band corresponds to the situation
where it is assumed that the CHIVE operation is required
to pay only the proportional page reader cost, based on
the fraction of the 8-hour shift that is actually
utilized for the CHIVE operation.
In all
possible is
page reader
cases, a text density
of 60% of the maximum
assumed, and it is further assumed that each
is operated at an effective speed de-rated
to 75% of its nominal capability. In the cases of both
the Rabinow and the Recognition Equipment systems, costing
was based on inclusion of 8000 words of memory in the
small computer associated with the page readers. Basic
page-reading operations can be achieved with 4000 words,
or even with 2000 words, but it is believed that at least
8000 words of memory are required in order to be able to
provide a reasonable amount of on-line pre-editing. In
the case of the Farrington machine, no computer memory is
provided--all programming and control are achieved by
means of a plug-board arrangement.
INPUT TRANSCRIPTION
Evaluations
7.B.2.4.
-315-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
ANNUAL COST (0
Approved For Release 2000/05/0?EW-RDP78-03952A000100070001-5
1,000,000
500,000
me??=0
I I II III I
I I
FARRINGTON
FULL COST
PROPORTIONAL COST
0 500,000
INDEX ITEMS/YEAR
OPERATING COSTS
FARRINGTON PAGE READING SYSTEM
FIGURE 73-6
1,000,000
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
1,000,000
ANNUAL COST (0
1
RABINOW (CDC)
FULL COST
PROPORTIONAL COST
1
500,000
INDEX ITEMS/YEAR
OPERATING COSTS
RABINOW (CDC) PAGE READING SYSTEM
FIGURE 7.8-7
1
1,000,000
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
1,000,000
500,000
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
RECOGNITION EQUIPMENT
FULL COST
PROPORTIONAL COST
500,000
INDEX ITEMS/YEAR
OPERATING COSTS
RECOGNITION EQUIPMENT PAGE READING SYSTEM
FIGURE 7.B-8
1,000,000
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/0SEMA-RDP78-03952A000100070001-5
The discontinuities of the upper cost bands of the
systems based on the Farrington and the Rabinow readers
(Figures 7.B-6 and 7) are caused by the fact that one
reader is not sufficiently fast to process as many as
one million items per year. Accordingly, the discontinuities
represent the added costs of an additional reader in each
of the two cases.
The correction time required for the page-reader
operations was specified as 1.0 minute per item, which is
well toward the lower end of the range for all of the
systems considered. This specification was based on the
assumptions that simple errors can easily be deleted
through the use of white over-paint (Snopake, etc.) or
adhesive strips in preparation for over-typing, and that
more extensive errors can be accommodated by computer-
programmed processing of coded correction entries at
the end of the item.
The costs indicated in Figures 7.B-6,7, and 8 were
all determined for off-line operation with single-font
page reader equipments. For the Recognition Equipment
reader, a configuration designed for on-line operation
with five fonts would cost approximately $8292 additional
INPUT TRANSCRIPTION
Evaluations
7.B.2.4.
-319-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08S.E6KIRDP78-03952A000100070001-5
per year.* No figure supplementing Figure 7.B-8 has been
made because this additional cost would not cause a
significant shift of the cost bands of that figure. The
greatest upward shift would be only approximately 5 percent
at the 100,000-items-per-year end of the upper band (full
cost support by the index operation), tapering down to
only a 1.5 percent increase at the 1,000,000-items-per-'year
end of that band. For the lower band (proportional
cost support by the CHIVE operation), the upward shift
would be only approximately 0.9 percent. However, the costs
for the on-line, five-font configuration of the Recognition
Equipment unit are included in Section 7.B.3. (Table 7.13-1).
It is not feasible at this time to provide graphs
of the costs that could be anticipated for systems based
on page readers procured from IBM RCA, GE, or any of
*This increase is due (a) to an added cost of $9492 per
year for the additional equipment required to increase
from 60 to 240 characters the number of patterns that can
be recognized, (b) to the added cost of approximately
$4,240 per year for interface equipment, and (c) to a
decreased cost of $5440 made possible by the elimination
of the requirement for the off-line magnetic tape transport
system. (It must be noted that the interface cost is that
specified for operation in conjunction with an IBM 7090
computer. Recognition Equipment does not ?as yet have
information for the IBM 360 computers on which to base cost
specifications).
INPUT TRANSCRIPTION
Evaluations
7.B.2.4.
-320-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0eW-RDP78-03952A000100070001-5
the other potential suppliers, because no reliable
specifications of the costs of their equipments have yet
been made available. It can only be stated that estimates
of the purchase cost of equipments obtained from these
manufacturers range from $500,000 to $1,000,000. For
this increased cost range, and using the upper band of
Figure 7.B-8 as a reference, an upward shift of approximately
$10,000 to $110,000 (per year) would occur.* The cost
increases for the lower curve (proportional support by
the indexing operation) would of course be a function of
the reading speed of the readers under consideration.
For the transcription, proofing, and correction of
from 100,000 to 1,000,000 index sheets per year, the
number of typists and proofing clerks required ranges
from 8 to 83. A complete tabulation is provided in
Section 7.B.3. (Table 7.B-2).
7.B.3. SUMMARY
The results of the cost evaluations of the four
major types of data input systems considered in this
*These shifts in the plot are based on a purchase cost of
approximately 8450,000 for the Recognition Equipment page
reader.
INPUT TRANSCRIPTION
Summary
7.3.
-321-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0?EMERDP78-03952A000100070001-5
study are combined graphically in Figure 7.B-9, and in
tabular form in Table 7.B-1 to facilitate direct
comparisons. Of prime significance is the fact that
with one exception,* the predicted costs of all systems
for processing of 1,000,000 indexing records per year
are of the same order of magnitude in the range from
approximately $440,000 to $750,000. This is due
primarily to two factors. First, the major cost for all
systems is the salaries of the typists (or keypunch
operators) and the proofing clerks. Consequently, the
influence of large variations in equipment costs is
subordinated--even though these costs differ in some
cases by an order of magnitude.** Secondly, the more
expensive equipments in general seem to be associated
with the faster, more efficient operations. Begause
of the preponderant salary costs, even a relatively
* The exception is the IBM 2250 Display Unit in the CRT/KB
category (Figure 7.8-4). As stated in Section 7.B.2.3.,
this equipment is more sophisticated than required for
the CHIVE indexing application, but it was nevertheless
included for reference purposes.
**For example, the Teleregister CRT/KB units cost approximately
$1,900 per unit per year, as compared to the Teletype
Model 33 paper tape punch at approximately $170 per unit
per year.
TiVr.:7
-322-
INPUT TRANSCRIPTION
Summary
7.B.3.
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
ANNUAL COST ($)
ANNUAL COST IV
500,000
INDEX ITEMS/YEAR
PUNCHED CARD SYSTEMS
1,000 000
500,000
INDEX ITEMS/YEAR
1,000 000
1,000,000
500,000
FARRINGTON
FULL COST
PROPORTIONAL COST
500,000 1,000 000
INDEX ITEMS/YEAR
PUNCHED PAPER TAPE SYSTEMS
500,000
INDEX ITEMS/YEAR
1,000000
ANNUAL COST ($)
1,000,000
500,000
1,000,000
.Z;
8
500,000
RABINOW (CDC)
FULL COST
PROPORTIONAL COST
0
500,000
INDEX ITEMS/YEAR
PAGE READING SYSTEMS
1,000000
FM,: IBM 2250
BURROUGHS
E7:1 TELEREGISTER 205 5
IPC 1560
500,000
INDEX ITEMS/YEAR
FIGURE 7.B-9 COMPARATIVE OPERATING COSTS OF CANDIDATE INPUT SYSTEMS (off-line)
1,000 000
ANNUAL COST ($)
1,000,000
500,000
o 0
0
04
CD -
a.
RECOGNITION EOUIPMIT
FULL COST 0
'S
PROPORTIONAL COST
CD
500,000
INDEX ITEMS/YEAR
CRT /KB SYSTEMS
500,000
INDEX ITEMS/YEAR
1,000000
Approved For Release 2000/05/0ga.baA7-78-03952A000100070001-5
COST SUMMARY
OF
CANDIDATE INPUT SYSTEMS
System
Yearly Cost (Thousands of Dollars)
100.000 Items/Yr.
500.000 Items/Yr.
1,000,000 Items/
Min.
Max.
Min.
Max.
Min.
Max.
I PUNCHED
CARDS
IBM 029/059
67.1
77.8
327
385
632
748
IBM 029/Listing
66.6
71.8
303
341
570
647
IBM 826
57.0
70.0
284
327
550
632
PUNCHED
PAPER_ TAPE
Dura Mach 10
58.7
70.5
286
322
549
621
Flexowriter
58.3
70.1
284
320
546
617
Greene Letteriter
58.7
70.5
286
322
-
549
621
Royal SDW 5000
60.8
73.2
296
334
370
645
Teletype #33
54.6
65.3
266
298
510
574
Teletype #35
57.0
68.3
277
312
533
601
CET/KB
*
Burroughs
90.5
105
259
297
481
549
dd 10
60.6
73.6
246
285
477
548
IBM 2250
97.3
129
485
565
956
1,110
IPC 1560
53.1
67.8
247
288
487
565
Philo? *
82.3
95.5
267
307
501
573
Telerigister 203
50.4
64.2
250
294
500
577
Off-line
single-font
JFull
Farrington*
Cost
95.4
101
247
279
488
546
Prop.
Cost
53.3
58.7
246
278
486
544
**
Rabinow
(CDC)
Full
Cost
91.3
96.7
243
275
479
538
Prop.
Cost
48.6
54.0
222
254
444
502
Recognition
'
Equipment
Full
Cost
170
175
321
353
516
574
Prop.
Cost
49'4
54.8
226
258
453
511
2 Recognition
ri g
I 'Equipment
0 1
CD u-
Full
Cost
178
184
330
361
524
583
P
Prop.
Cost
49.9
55.2
228
260
457
515
* *
Not a production item -- pricing is approximate.
Nibmpined FamRe1eaure20010/05/08 4C144116Q78-03952A000100070001-5
For Farrington system--at more than 510,000 items/yr.
For Rabinow system--at more than 870,000 items/yr.
Approved For Release 2000/05/085:ECUIRDP78-03952A000100070001-5
small advantage in speed of transcription or error
correction can offset a substantial disadvantage in
equipment costs.
The uniformity in costs is not so pronounced in all
cases for lower production rates, particularly at 100,000
indexing records per year. This non-uniformity is due
to relatively large, "non-modular" equipment costs which
are not averaged out significantly at the lower production
rates. The most significant cases of this type are the
systems based on Burroughs and Philco CRT/KB units, and
the systems based on page readers where full(equipment)
costs are assumed to be borne by the indexing operation.
The number of typists (or keypunch operators) and
proofing clerks required for the various systems are
tabulated in Table 7.8-2. For the extreme cases, the
predicted numbers range from 70 to 129 typists and clerks
for production of 1,000,000 indexing records per year.
These numbers represent very substantial staffing
problems. It is recommended that more extensive and
continuing in-house tests be made under conditions as
representative of anticipated operating conditions as
possible to validate the initial timing assumptions and
to uncover ways to improve the operation.
INPUT TRANSCRIPTION
Summary
7.B.3.
-325-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Table 7.B-2
SUMMARY OF REQUIREMENTS FOR TYPISTS/KEYPUNCH
OPERATORS AND PROOFING CLERKS FOR CANDIDATE
INPUT SYSTEMS
SYSTEM
Number of Typists (or Keypunch Operators) &Proofing Clerks
100 000 Itens/Yr
500,000 Items/Yr.
1,000,000 Items/Yr.
Min.
Max.
n.
ax.
in.
' ',3c.
i3UNCHED CARDS
IBM 029/059
11
13
54
65
107
,
129
IBM 029/Listing
10
11
47
54
94
108
IBM 826
8
10
40
47
80
93
Punched Paper Tape
9
11
46
52
91
103
, CRT/KB
7
9
35
41
70
81
Page Reader
8
9
36
42
72
83
,
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
11111111111111 1
Approved For Release 2000/05/0SEGZIRIE5f-RDP78-03952A000100070001-5
Many measures might be used to make a relative
comparison of the four approaches chosen for this analysis.
Some of these measures are noted in Table 7.B-3. Early
in the analysis it was suggested that cost was the main
factor, with staff size another important consideration.
For this reason, most of the attention was directed to
these two factors.
In any event, since the personnel costs are
preponderant, and since providing working space and
facilities for the personnel also constitutes a significant
problem, all reasonable efforts should be expended to
improve operating efficiency in order to minimize the
personnel requirements. In this connection, the decision
by the OCS and the CHIVE staffs to plan for use of a
page-reading system offers the possibility of some gains
in this direction with the hope of processing many or all
of the indexers' work sheets directly on the page reader,
thus eliminating the transcription process in part, or
perhaps even completely (c.f., Section 7.B.1.). If
feasible to any significant extent, this could conceivably
result in very appreciable reductions in costs, space,
and manpower requirements.
INPUT TRANSCRIPTION
Summary
7.B.3.
-327-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Factor
Table 7.B-3
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Comparison of Alternative Input Transcription Methods
Punched Punched
Cards Paper Tape
CRT/KB
Optical
Page Reader
Cost for Final System (annual cost 550-748,000 510-645,000 477-577,000 457-583,000
in dollars)*
Staff for Final System** 80-129 91-103 70-81 72-83
Floor Space for Final System (sq.ft.)*** 3645-6210 3640-4160 2800-3240 3180-3660
Time Lag to get final indexing record
into computer system Moderate Moderate Very small Small
Ease of Error Notation & Correction Moderate Moderate Very good Good
Compatibility with current staff,
equipment, & procedures Good Some Difference Some Difference Some Difference
Possible Maintenance problems Considerable Maint. experience exists Relatively little maint. experiencE
for these equipments. exists for these equipments
Space required for Materials Storage
(in addition to space noted above) Some Some None Some
cncn
rn w Need Special Controls on print quality None None None Definitely nn
73 3 Training Required by Transcription m
-1 Clerk About the same for all systems ?I
Capability for Immediate Recall of
Earlier Records for Review or
Manipulation Moderate Moderate Very good Good
Possibility of Indexers Eventually
Entering Information without Paper
Transcription Poor Poor Very good Good
Special Problems Display limited to 500-
1000 char, at one time.
Possible flicker or reso-
lution (human factors)
problems.
* See Table I in CHIVE R-2-65
** See Table II
*** See Table III
401
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
1 1 1 1 1 1
Approved For Release 2000/05/0?K:RgRDP78-03952A000100070001-5
Planning to use a page reader suggests a number of
follow-on tasks. An indication of the activities that
such a decision dictates is provided by the list in
Table 7.B-4. One of the principal tasks noted is that
of choosing a specific page reader model. Some tentative
specifications are set forth in Table 7.B-5. At the
time these specifications are completely detailed, other
models than those herein considered may appear as likely
candidates.
INPUT TRANSCRIPTION
Summary
7.B.3.
-329-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release320001CIE500tE:C8ff-RDP78-03952A000100070001-5
Subsequent Tasks for OCR and OCS After Choosing the Page Reader Approach
(not necessarily in sequential order)
A. Make arrangements with other potential users to share the page reader facility.
1. Agreement on type fonts (choice of specific fonts, and their sizes, and
the number of characters required).
2. Agreement on location of equipment (convenience, power, refrigeration,
noise, structural considerations).
3. Initial time scheduling.
4. Possible need by other users for peripheral page reader features
(off-line magnetic tape transport, printer, multiple page hoppers).
5. Initiate study of other OCR application's within broad CHIVE context
(e.g., reading library requests or acquisition forms).
B. Establish the Basic System Specification (see Table 7.B-5).
1. Equipment.
2. Environment.
3. Delivery and maintenance.
C. Choose the specific page render model.
D. Establish Document Format.
1. Forms design.
2. Determination of forms suppliers.
3. Procurement and test of forms.
E. Establish Control of Typing Quality.
1. Specifi.cation, selection, and procurement of special typewriters.
2. Specification of materials (paper, ribbons, correction material).
F. Establish Tolerable Error Levels.
1. Determine relative criticality of various elements of the index record.
2. Determine what error checking or proofing procedures (e.g., use of check
digits or table look-up verification for selected elements) are reasonable.
G.
Prepare instruction manual for typists.
H. Perform computer programming of the page reader control functions.
Approved For Release 200149?L08 : CIA-RDP78-03952A000100070001-5
SECRET
SECRET
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Table 7.B-5
TENTATIVE PAGE READER SPECIFICATIONS - PROJECT CHIVE
A. Fonts
1. Need a capability of handling two or more fonts
with 60-character sets as follows:
a. A standard IBM electric typewriter font
(Elite), 10 or 12 character per inch
and 6 or 8 lines per inch. Upper case
alphabetic characters will be used.
b. One NPL font compatible with the IBM 1403
printer in size and spacing; up to 132
characters per line.
2. Require a reader which can handle one font
per page and can change from one font to
another without manual intervention (a special
symbol may be employed to signal a change in
font).
B. Reading Speeds
1. Reading speeds can be specified only for
throughput for the system. Eventual requirement
is for 1,000,000 records of 1000 characters each
plus special query requirements. This volume is
to be read in a normal year of 52 weeks, 40 hours
per week. This requires a system reading speed
of 150 or more characters per second throughput
with a page throughput of 10-15 pages per minute;
each page to contain an average of 1000 characters.
This is a total system capability--one or more
readers.
-331-
Approved For Release 2000/05/08sLERTRDP78-03952A000100070001-5
Approved For Release 2000/05/0?E:C8R-RDP78-03952A000100070001-5
C. General Features
1. Page sizes possible:
a. Normal page sizes from 3 x 5 to 8.5 x 14 inches
b. Normal IBM 1403 printer paper - 10 x 14 inches
(up to 132 char/line).
2. Possibility of intermixing different size pages
in the input hopper.
3. Need for some means of segregating rejected pages
from error-free pages in the output stacking
station (or stations).
4. Input stacking station should have a capacity
of 50 or more pages with a capability of adding
to the stack while the reader is running. Output
stacking station(s) should have a capacity of at
least 200 sheets contingent on the relative
position of the input and output stacking stations.
5. Need a capability of marking rejected lines on the
rejected page.
6. Equipment should be capable of ignoring at least
one background color so that pre-lined forms may
be used.
7. The reader should be sensitive to mark sensing.
8. Capability for handling normal busahess paper stock.
D. Characteristics of Page Reader's Control Processor--
If System Acquired has a Control Processor
1. Need 16,000 or more 8 bit characters of core
storage or the equivalent
2. Interfaced for on-line operation with IBM 360 with
a controlled interrupt capability (parity and
character length compatability is implied).
-332-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0gWRDP78-03952A000100070001-5
3. Ability on the part of the processor to signal the
system to mark the lines (and offset stack rejected
document) or give the line and character count
of an unrecognizable character.
4. Ability to send documents to related output
stacker(s) under program control.
5. Processor should be able to signal the equipment
to rescan a line.
6. Ability to do some simple formatting and editing
of information from the document before it is
transferred to the IBM 360.
7. Software package for the control processor should
be compatible with Operating System/360.
E. Operation Guarantees
1. Delivery Schedule. Delivery required to interface
with the IBM 360 installation schedule of February
April 1966.
2. Maintenance
a. If equipment purchased:
Either (1) complete maintenance training
for CIA employees and on-call
(4 hrs.) engineering service,
or (2) contract maintenance including
Eli on-call (4' hrs.) engineering
maintenance.
b. If equipment rented. Full preventative and
engineering maintenance.
010 3. Performance. No more than 10 rejects per 100,000
input characters from normally prepared cleanly
typed documents. No more than 10 characters
substituted per 100,000 input characters from
normally prepared cleanly typed documents.
NINO -333-
mei
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08SaATRDP78-03952A000100070001-5
F. Environmental Conditions
To meet normal government specifications.
G. Spare Equipment Required
1. If purchased. Recommended list of spare parts and
costs (unless provided for in maintenance contract)
derived from a mean-time-to-failure study of
equipment components.
2. If rented. Sufficient spare parts to allow
normal maintenance replacement within 4 hours.
-334-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08sMATRDP78-03952A000100070001-5
Appendix 7.C.
EDP OUTPUT FORMATS
The standing report definitions which will be
provided initially have been enumerated in this appendix.
In order to demonstrate the makeup of these reports,
the layouts of representative types have been
prepared.
7.C.1. DOCUMENT INDEX FILE REPORTS
Figure 7.C-1, Document Identification and Term
Report, consisting of three parts, shows the type of
report which may be generated as the result of a
search of the Document Index Record file.
The cover page, which may actually be a back cover
since it contains information extracted from each of
the records in the reported file, lists the document
access number for all document reports, in the order of
their selection. As a summary sheet it can be used by
the requestor to demarcate all of those documents
which he wishes retrieved for more detailed examination.
The first page presents, for subsequent reference,
the specifications upon which the search was based. The
OUTPUT FORMATS
File Reports
7.C.1.
-335-
Approved For Release 2000/05/08 i.R.O,E1DP78-03952A000100070001-5
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
cn
rn
(*)
X:1
rn
DATE
Figure 7.C-1
DOCUMENT ID-TERM REPORT-COVER PAGE
*.* SECURITY CLASSIFICATION * *
COVER
DOCUMENT INDEX SEARCH PAGE
I
REQUESTOR NAME:
* * *
REQUESTOR LOCATION:
REQUEST NUMBER:
DOCUMENTS SATISFYING THIS REQUEST:
* * SECURITY CLASSIFICATION * *
COVER
PAGE
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
1""11.1
1
I t 1 1. 4 1 L t I t tit ta.t
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
DATE
REQUEST NUMBER
SEARCH LOGIC:
Figure LC-1
DOCUMENT ID - TERM REPORT - FIRST PAGE
* * SECURITY CLASSIFICATION * *
DOCUMENT INDEX SEARCH
* * *
JOB RETRIEVE:
TESTS:
PROCESS:
En JCL:
X
PAGE 1
X
* * SPCUITJ:TY CLAZSDICATIOU * *
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
FIRST
PAGE
DATE
cn i
rn c...)
CI to
m co X: x
ACCESSION
NUMBER
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SERIES ID
NUMBER
Figure 7.C-1
DOCUMT ID - TERM mow: - BODE PAGE
* * =MITE CLASSIFICATION * *
DOCUMENT INDEX SEARCH
* * *
77711
TERMS
(TAG: VALUE)
PAGE 2-N
x x x x
x x x x x
x x x
* * SECURITY CLASSIFICATION * *
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
DOC ID N
TERMS
REPORT
CJI
I1
rn
Approved For Release 2000/05/0WM-RDP78-03952A000100070001-5
remaining pages highlight the access number and
identification of each of the documents whose index
records meet the retrieval criteria and it then
presents selected terms appearing in that index record.
Figure 7.C-2, Abbreviated Header and Contents
Report, will use the cover and first pages as they
appear in 7.C-1 and the content pages as they appear in
7.C-2. The content pages will contain selected items
out of the document index record header phrase and
the entire content of the remainder of that record.
Figure 7.C-3, Full Header and Contents Report
will use the cover and first pages as they appear in
7.C-1 and the content pages as they appear in 7.C-3.
The entire content of both the header phrase and the
remainder of the document index record will be printed.
Figure 7.C-4, KWIC - Document Index Titles will
also use the cover and first pages as shown in 7.C-1.
The remainder of the report will present only the
document accession number and a fixed number N of
characters of the document title. The whole title will
be permuted cyclically so that a key word appears, left-
justified, at the left double column shown in the figure,
and the N characters which result from that permutation
OUTPUT FORMATS
File Reports
-339-
Approved For Release 2000/05/08ER.FDP78-03952A000100070001-5
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Figure 7.C-2
ABBREVIATED HEADER - CONTENTS REPORT
* * SECURITY CLASSIFICATION * *
DATE ? DOCUMENT INDEX SEARCH
ACCESSION DISSEMINATION
DIMMER
TITLE: x
CONTROL
* * *
PAGE 24,N
TYPE CLASS DOCUMENT REPORT PUBLICATION INFORMATION
CODE COLE SERIES NO SOURCE DATE DATE
X XX X X X X X X X X X
fll
nl
X
PHRASE 1: x
PHRASE 2: x
PHRASE N: x
* * SECURITY CLASSIFICATION * *
ABBR HEADER N
CONTENTS
REPORT
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
11311111111111iIi
LI i I iiititii
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
DATE
ACCESSION DISSEMINATION TYPE
NUMBER CONTROL CODE
EVALUATION
Figure 7.C-3
FULL HEADER - CONTENTS REPORT
* SECURITY CLASSIFICATION * *
DOCUMENT INDKK SEARCH PAGE 24N
CLASS Documgal REPORT PUBLICATION INFORMATION
CODE SERIES NG SOURCE DATE DATE
or)
TOTAL ITEM VOI, ED ISSUE COUNTRIES CROSS CITED SENDER RECEIVER rn
r)
PAGES PAGES-REF REF . LOC LOC ? 2,
rn
-4
X X X X X X X X X X
LANGUAGE
TITLE:
TRANSLATEIG
AGENCY
PUBLISJER
x x x x x x x
EDITOR
PUBLISHER
LOCATION
PHRASE x
PHRASE N:
x
FULL HEADER N
CONTENTS
REPORT
* * SECURITY CLASSIFICATION * *
Approved For Release 2D07-0510137CW-RDP78-03952A000100070001-5
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
DATE
Figure 7.0-4
KWIC TITLE REPORT
* * SECURITY CLASSIFICATION * *
KWIC - DOCUMENT INDEX TITLES PAGE
TITLE
* * SECURITY CLASSIFICATION * *
KWIC
TITTLE
ACCESSION NO
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
111311MM
Approved For Release 2000/05/0*cM-RDP78-03952A000100070001-5
will be printed. Titles longer than N characters may be
truncated at both ends and those shorter than N will be
padded out to N characters with blanks between the
original ending and beginning characters of the title.
There will be as many recurrent permutations of the
title appearing as lines as there are key words in the
title. Normally the lines will be sorted on the basis of
the characters appearing between the two double columns
shown in the diagram, which for sorting purposes will
be treated as a single key field.
Figure 7.C-5, Map Catalog Card, will be printed
on a fanfold of 3 x 5 inch cards for use by the Map
Library. Condensed cover and terminating cards may be
specified similar to 7.C-1 but the content pages will
appear as in 7.C-5.
7.C.2. VOCABULARY CONTROL FILE REPORTS
Figure 7.C-6 is an example of the type of report
which may be prepared to present vocabulary control
information about organization and facilities. The
information retained in the Master Organization and
Facility Information File will be printed a record at
a time. Where such information has been inserted in the
OUTPUT FORMATS
Vocabulary Control
7.c.2.
-343-
Approved For Release 2000/05/08 ip&IDP78-03952A000100070001-5
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Figure 7.C-5
MAP CATALOG CARD
* AMA *
TITLE
* SCALE * * CLASSIF * * ML AREA CODE *
* SUBJECT * * SUBJECT NO * * ACC NO *
DATE.
PUBLISHER
SOURCE
EDITION
PUB NO
* DESCRIPTION *
ADDED ENTRIES
MAP CATALOG CARD
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
1 1111IIIIIII
I I I I I I I I L L I I. LI
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
DATE
x<
Figure 7.C-6
ORGANIZATION/FACILITY CONTROL REPORT
0 ASSIGNED NAME:
0 FOREIGN NINE:
0 VARIANT NAME:
0 PREVIOUS NAME:
o)
rn 0 NAME ABBR:
rn
-4 TEL CODE: x
CR NO'S:
ADDRESS:
* * SECURITY CLASSiFICATION * *
ORT ZATION/FACILITY CONTROL REPORT
ORGANIZATION NAME -----
PAGE
FIB- x
PART ORG:
0 REMARKS:
x, TYPE 0/F:
x, ra x
x, SR- x x, COMOR- x x, BR- x
x, BE- x x, NPIC- x
SOURCE CITATIONS:
Approved For Releatd'2DiabigagWatqW4-te952A000100070001-5
Approved For Release 2ooOreme : CIA-RDP78-03952A000100070001-5
file, cross referencing of sources for the data printed
will be listed in summary for each record.
7.C.3. SUMMARY DATA FILE REPORTS
Figure 7.C-7 presents the cover page for a
Leader Appearance Report, describing the basis for the
retrieval of the information reported, and a content
page which includes for each leader cited the cross
references of the sources with the information listed.
Figure 7.C-8, which will be accompanied by a
cover page such as appears in 7.C-7, shows the content
pages for such a report where no cross referencing
information about sources is provided.
7.C.4. UNSYNTHESIZED FILE REPORT
Figure 7.C-9, Foreign Travel Report, shows a cover
page describing the basis for the retrieval of records
for the file reported, and the content pages showing
selected information about personalities who have been
reported to have travelled to places of interest. Since
the information is unsynthesized and unevaluated all
sources used will be cited. The report may contain
redundant, or incorrect information but can be used in
developing evaluated items about the travel of these
individuals.
OUTPUT FORMATS
Unsynthesized File
7.C.4.
-346-
Approved For Release 202E1M8 : CIA-RDP78-03952A000100070001-5
I t II I I tit t
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
DATE
Figure 7.C-7
LEADER APPEARANCE - SOURCE CITATION REPORT - COVER PAGE
* * SECURITY CLASSI.b.LCATION * *
LEADER APPEARANCE BORT
* * *
PAGE COVER
REQUESTOR NAME PERIOD OF REPORT
REQUESTOR LOCATION
REQUEST NUMBER
0 1
rn
(-1 w
PRIMARY SEARCH CRITERIA
rn
-4 1
SEARCH LOGIC:
JOB RETRIEVE: x
TEST:
PROCESS:
END JOB:
x TO x
COVER
PAGE
* * SECURITY CLASSIFICATION * *
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
DATE
NAME
SOURCE:
Figure 7.C-7
LEADER APPEARANCE - SOURCE CITATION REPORT - BODY PAGE
* SECURITY CLASSIFICATION * *
LEADER APPEARANCE REPORT
TITLE CITIZENSHIP APPEARANCE
X X
x
SOURCE: Ox
X X
"1
SOURCE:
PAGE 1-N
DATE CITY COUNTRY FUNCTION
x xO.
4110'
.
X X
X X
X X
X X
x
(i)
rn
rn
X X
X X
X X
x
X X
X X
X X
x
X X
X X
x x
X X
x x
X X
* * SECURITY CLASSIFICATION * *
LEADER
APPEARANCE
REPORT
WITH SOURCES
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5 _
1 ?11
t I &tilt t
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Figure 7.C-8
JO:E7,,SEL LEADER APPMARAiCE REPORT
* * E5R1T CLASSLICATION * *
LEADER APPEARANCE REPORT
DATE
CITIZENSHIP
X X
X X
PAGE -4N
APPEARANCE
DATE CITY . COUNTRY FUNCTION
X X XX X X X X
X X X X X X
X X X X X X
X X
X X
XX
X X
X X
Xx XX
X X X X
X X
X X
LEADER
APPEARANCE
REPORT
TAT/0 SOURCES
* * =RIM CLASSIFICATION * *
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Figure T.C-9
INDIVIDUAL TRAVEL TO FOREIGN COUNTRIES REPORT - COVER PAGE
* * SECURITY CLASSIFICATION * *
DATE? FOREIGN TRAVEL REPORT
* * *
REQUESTOR NAME
REQUESTOR LOCATION
rn
REQUEST NUMBER
() LTI
7, 0
rn 1 PRIMARY SEARCH CRITERIA
-.1
SEARCH LOGIC:
,[6,[ L.1111 ,[1,f1111 41,4. kit
JOB RETRIEVE: x
TEST:
PROCESS:
END JOB:
PAGE COVER
* * SECURITY CLASSIFICATION * *
COVER
PAGE
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
1 1111111M
I_
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
DATE
Figure 7.C-9
INDIVIDUAL TIAVM TO FOREIGN COUNTRIES REPORT - BODY PAGE
* * SECURITY CLASSIFICATION * *
FOREIGN TRAVEL REPORT
PERSONAL= OCCUPATION
* * *
PAGE
DATE OF TVL REASON TRAVEL TO SOURCE
x X X X X X X X X X X x
x X X x x x x x
0 '
m (-0x
n ui x XX XX XX
xi H
m 1
--1 x x x x x , x x x x x x x
x x
X X X X X X X X
X X X X X X X X X X
X X X X X X
X X X X X X
X X
X X x x X X X X
* SECURITY CLASSIFICATIO,
* *
INDIVIDUALS N
TRAVEL
REPORT
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/OPW-RDP78-03952A000100070001-5
Appendix 7.D.
SYSTEM MACROS USED IN OPERATIONAL PROGRAMS
This section is an introduction to many of the macros
of OS/360 that are used by the operational programs. It
gives the name and a brief description of the macro, but
assumes the reader has read Chapter 7.5. of this report.
The departure macros are used when the processing of
a task passes from one operational program to another
operational program. The details for these macros are
discussed in CHIVE/N-8-64.
Data set control macros are used to find and keep track
of data sets as well as to open and close data sets. Data
set accessing macros are used for reading and writing records
of data sets. These macros are discussed in Section 7.5.7.
of this .eport and in CHIVE/N-1-65.
Storage allocation macros are used to request and release
core storage. These are also discussed in Section 7.5.6.
of this report and in CHIVE/N-9-64.
Preliminary versions of the macros in this report
are discussed in CHIVE/N-5-64.
SYSTEM MACROS
7.D.
-353--
Approved For Release 2000/05/08s:MiRDP78-03952A000100070001-5
Approved For Release 2M6A%T08 : CIA-RDP78-03952A000100070001-5
7.D.1. DEPARTURE MACROS
Departure macros vary depending on the task structure
used. The task structure can be:
- Simple: one load module (a set of programs always
loaded together) which contains all the code
required for task execution.
- Overlay: similar to above except that a preplanned
overlay pattern in the load module is followed so
that all the code required at any time is in core.
- Dynamic serial: more than one load module is
called during the course of program execution.
Transfers between load modules go through the
Executive Control Program.
- Dynamic Parallel: an asynchronously operating sub-
program (lower level program) is established as a
task and follows normal task rules.
7.D.1.1. Direct Transfers (CALL)
Direct transfers are transfers between programs in
the same load module. The linkage editor (part of the
program loader) sets up the transfers when the programs
in the load module are loaded.
The direct transfer macro is "CALL, A" where A is
the name of the subprogram being transferred to. When
subprogram A desires to transfer back to the calling
program it does so by means of a "RETURN" macro. Neither
transfer goes through the Executive Control Program. Methods
SYSTEM MACROS
Departure Macros
7.D.1.1.
-354-
Approved For Release 2W9N08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/0V%U-RDP78-03952A000100070001-5
of passing parameters can include linkage in specified
registers or bytes relative to the "Call A" macro.
In the overlay structure transfers are the same except
that a segment load "SEGLD" macro is given to start the
loading of a segment and segment wait macro "SEGWT" is
given to insure the segment has been loaded before
direct transfers to the segment are started.
7.D.1.2. Indirect Transfers Within the Same Task
These transfers are used when transferring to a
program in a different load module in the same task.
These transfers must go through the Executive Control
Program.
7.D.1.2.1. LINK Macro
"LINK" is used to pass control from one load module
to another when a "RETURN" is expected. Both the "LINK"
and HRETu q" go through the Executive Control Program.
The form of the macro is "LINK, A" where A is
the symbolic name of a lower level program in a different
load module that is being transferred to. Parameters
cannot be passed relative to the "LINK" macro.
7.D.1.2.2. XCTL Macro
"XCTL" is the macro used to pass control from one
load module to another when the program being transferred
SYSTEM MACROS
Departure Macros
7.D.1.2.2.
-355-
Approved For Release 2000/05/ONEWDP78-03952A000100070001-5
Approved For Release 201f6/RT08 : CIA-RDP78-03952A000100070001-5
to is on the same level. This macro is the same as "LINK"
except that A is on the same level and no "RETURN" is
given to the calling program.
7.D.1.2.3. LOAD Macro
"LOAD" can be used instead of "LINK" to bring a
program into core and then the transfer and return can be
direct. "LOAD" is used when a task must make frequent
use of the same load module. The "LOAD, A" macro performs
the function of finding program A, allocating space,
fetching, and telling the calling program where A is
located, but not transferring to A. Program A is placed
in a reserved status so that it will not be overlaid. The
reserved status can be cancelled by a "DELETE, A" macro.
After A is loaded, transfers between it and the
calling program are direct by means of a BALR instruction
and a "RETURN" macro.
7.D.1.3. Indirect Transfer for Parallel Processing (ATTACH)
The "ATTACH" macro creates a new task in the system
that can proceed in parallel with the creating task.
"ATTACH" should be used only when a significant amount of
overlap between the two tasks can be achieved. The
parent task (the task giving the ATTACH) can give the
SYSTEM MACROS
Departure Macros
7.D.1.3.
-356-
Approved For Release 2g08&08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/fficM-RDP78-03952A000100070001-5
sub-task (the task created by the ATTACH) a higher
or lower priority than its own to help create an
advantageous overlap of operations. The parent task can
check for completion of the sub-task with the "WAIT" macro.
7.D.1.4. Other Departure Macros
Some departure macros not covered above involve a
task terminating itself and a task waiting for completion
of IO or another task. A task can normally or abnormally
terminate itself. When a task terminates itself, it also
terminates all its subtasks, etc.
When a task cannot proceed further before completion
of some order it has given (ID command or subtask),
It issues a "WAIT". The Executive Control Program will
not return CPU control to this task until its "WAIT" is
satisfied.
7.D.2. DATA SET CONTROL MACROS
The "OPEN" macro is used to get a data set ready
for accessing. This must be done before any data is
read or written. The only data sets which a program
does not need to "OPEN" are supervisor controlled data
sets such as the catalog, the system library, and the
data sets used for roll-out. These data sets are referred
to indirectly by the problem program.
SYSTEM MACROS
Data Set Control
1,D.2.
-357-
Approved For Release 2000/05/ONEWDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
"OPEN" will fill in all DCB (Data Control Block)
fields and establish the interconnections to the supervisor
programs that will handle all IO for this data set.
These interconnections included buffering and accessing
routines built to handle requests on this data set.
The "CLOSE" macro restores the DCB to its original
form prior to "OPEN" and frees the storage of the various
lists and routines built up by the supervisor to handle
this data set.
7.0.3. DATA SET ACCESSING MACROS
When a program opens a data set, it specifies the
accessing method it will use. The OPEN macro sets up
accessing routines and buffering routines to handle this
accessing method. These accessing methods are discussed
in Section 7.5.7. The user's request for IO goes to
these accessing and buffering routines, which communicate
the need for 10 to the supervisor by a system macro that
references a prewritten channel program. The types of
channel programs which the OPEN will set up for the
following accessing methods are listed below.
Accessing Method Macros
BSAM Read, Write
QSAM Get, Put
SYSTEM MACROS
Data Set Accessing
7.D.3.
-358-
Approved For Release 2et4fig08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/NR:E&A-RDP78-03952A000100070001-5
BISAM Read, Write
QISAM Setl, Get, Put
BPAM Read, Write, Find, Stow, Bldl
- FIND: initializes the sequential accessing of a
particular named member of the data set. It
locates the member desired. Subsequent READ
macros retrieve the physical records that make
up the member.
- STOW: is used to change the directory entries of
the data set.
- BLDL: reads control information from the directory
based on a list of names. This control information
can be saved in core so subsequent FIND macros can
use it.
- READ: for BISAM is a request to read a physical
record with the key specified from the data set.
- READ: for other access methods is a request for the
next physical record in the data set.
- WRITE: for BISAM is a request for a physical record
to be written on a data set. The key specified will
determine the order of the record in the data set.
WRITE: for other access methods is a request for
a pnysical record to written in sequence on the
data set.
- GET: is the request for a logical record, it
retrieves the next record in sequence from the
data set.
- PUT: is the request to write a logical record as
the next in sequence logical record of the data set.
- SETL: is the macro that specifies a new starting
point for accessing a data set in sequence.
SYSTEM MACROS
Data Set Accessing
7.D.3.
-359-
Approved For Release 2000/05/CAtRft-RDP78-03952A000100070001-5
Approved For Release 20W8T08 : CIA-RDP78-03952A000100070001-5
7.D.4. STORAGE ALLOCATION MACROS
7.D.4.1. GET MAIN Macro
GET MAIN is the operational programmer's macro
request for core storage. The amounts of space allocated
by type of request in OS/360 are:
One Block
Request
Multiple Block
Request
Non-Conditional Block of size Each block of a
must have specified specified size
specified amount
Conditional - Block of size Each block of a
Special return specified specified size
if request
cannot be met
Variable -
Special return
if minimum
length is not
available
Largest block
available within
specified limit
7.D.4.2. FREEMAIN Macro
FREEMAIN is the macro instruction to dynamically
release main storage so that it becomes available to
system for future use. The return procedure is
checked for recognizable errors. If the return frees a
2K byte block, it is returned to overall free storage.
A check is made to see if the 2K block could fulfill
pending roll-out or is needed for roll-in.
SYSTEM MACROS
Storage Allocation
7.D.4.2.
-360-
Approved For Release 2g9igf/08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/01FM-RDP78-03952A000100070001-5
Appendix 7.E.
HARDWARE UNIT CHARACTERISTICS
In this appendix the hardware characteristics are
divided into CPU, storage, and IO hardware characteristics.
The TO hardware characteristics are further divided into
channels, bulk storage I/O devices, and I/O devices used
for communication between the user and the system.
7.E.10 CENTRAL PROCESSING UNIT
The major parts of the /360 CPU which the user is
interested in and the difference between the various
models of the /360 are discussed in this section.
70E01.1. Characteristics That Differ Between Models
The difference between the models of the /360 is
storage capacity and speed, and the amount of I/O which
the modui. storage speed can support. Another difference
is the speed of the circuits.
These differences in speed and I/O capability greatly
affect over-all performance; none affects the appearance of
the design to the programmer, nor does it affect the results
of any program that does not depend on time.
Capacity of main core storage varies from 8,192
to 524,288 bytes, depending on the system model. A
HARDWARE CHARACTERISTICS
Central Processing Unit
7.E.1.1.
-361-
Approved For Release 2000/05/08s:EekiiRDP78-03952A000100070001-5
Approved For Release 20V6113g08 : CIA-RDP78-03952A000100070001-5
byte consists of eight bits plus a parity (check) bit,
and is the smallest addressable unit in the IBM System/360.
Storage capacities are always given in bytes. The main
storage capacity, speed, and access width for the models
of the 360 is shown below.
Main Storage
Capacity
(bytes)
System Model
8,192 C30
16,384 D30 D40
32,768 E30 E40
65,536 F30 F40 F50
131,072 G40 G50 G60
262,144
524,288
H40 H50 H60 H62 H70
160 162 170
Bytes per Access 1 2 4 8 8 8
Main Storage Speed 2 us 2.5 us 2 us 2 us 1 us 1 us
A channel directs the flow of information between
input/output devices and main storage. It relieves the
central processing unit of the task of communicating
directly with input/output devices and permits processing
of data to proceed concurrently with input/output devices.
In general, I/O operations on a channel are overlapped
with processing, and all channels can operate simultaneously,
provided only that the processing unit's data rate capa-
bilities are not exceeded. Nominal data rates for the
HARDWARE CHARACTERISTICS
Central Processing Unit
7.E.1.1.
-362-
Approved For Release 219,Jt115/08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/669M-RDP78-03952A000100070001-5
selector channels range from 250 thousand bytes to 1.3
million bytes per second, depending on the system model
and the channel options selected.
The only exception to the foregoing is a special
high-speed selector channel available for Model 50,
which operates in non-overlap mode.
The number of selector channels available on the
various models and their maximum data rates are shown
in the following table.
/360 Model
30 40 50 60 62 70
Channels
Maximum number 2 2 3 6 6 6
Sel Ch #1 (250 KC) X X
Sel Ch #2 (250 KC) X X
Sel Ch #1 (400 KC) X X
Sel Ch #2 (400 KC) X X
Sel Ch #3 (400 KC) X
Sel Ch #1 (1.3 MC) X X X X
Sel Ch #2,3,4,5,6
J.3 MC) X X X
A multiplexor channel can be substituted for one
of the selector channels.
7.E.1.2. 16 General Registers
The CPU can address information in 16 general registers.
The general registers can be used as index registers, in
HARDWARE CHARACTERISTICS
Central Processing Unit
7.E.1.2.
-363-
Approved For Release 2000/05/0%iagyRDP78-03952A000100070001-5
Approved For Release 20%tran8 : CIA-RDP78-03952A000100070001-5
address arithmetic and indexing, and as accumulators
in fixed-point arithemtic and logical operations. The
registers have a capacity of one word (32 bits). The
general registers are identified by numbers 0-15 and are
selected by a four-bit field in the instruction called
the R field.
For some operations, two adjacent registers can
be coupled together, providing a two-word capacity.
In these operations, the addressed register contains the
high-order operand bits and must have an even address,
while the implied register, containing the low-order
operand bits, has the next higher address.
7.E.1.3. 4 Floating-Point Registers
Four floating-point registers are available for
floating-point operations. These registers are two words
(64 bits) in length and can contain either a short (one
word) or a long (two word) floating-point operand. A
short operand occupies the high-order bits of a floating-
point register. The low-order portion of the register
is ignored and remains unchanged in short-precision
arithmetic. The floating-point registers are identified
by the numbers 0, 2, 4, and 6. The operation code
determines which type of register is to be used in an
operation.
-364-
HARDWARE CHARACTERISTICS
Central Processing Unit
7.E.1.3.
Approved For Release 220849/08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/01961k-RDP78-03952A000100070001-5
7.E.1.4, Instruction Sets
System/360 is a general-purpose system designed to
be tailored for commercial, scientific, communications,
or control applications. A Standard instruction set
provides the basic computing function of the system.
To this set a decimal feature may be added to provide a
commercial instruction set or a floating-point feature
may be added to provide a scientific instruction set.
When the storage protection feature is added to the
commercial and scientific features, a Universal set is
obtained.
7.E.1.5. Types of Arithmetic--Fixed Point, Decimal,
Floatiaa. Point
Fixed Point
The fixed-point instruction set performs binary
arithmetic on operands serving as addresses, index
quantitit and counts, as well as fixed-point data. In
general, both operands are signed and 32 bits long.
Negative quantities are held in two's-complement form.
One operand is always in one of the 16 general registers;
the other operand may be in main storage or in a general
register.
The instruction set provides for loading, adding,
subtracting, comparing, multiplying, dividing, and
HARDWARE CHARACTERISTICS
Central Processing Unit
7.E.1.5.
-365-
Approved For Release 2000/05/08s:EabRDP78-03952A000100070001-5
Approved For Release 20NanTh8 : CIA-RDP78-03952A000100070001-5
storing, as well as for the sign control, radix conversion,
and shifting of fixed-point operands. The entire instruction
set is included in the standard instruction set.
The condition code is set as a result of all sign-control
add, subtract, compare, and shift operations, and can be
checked by logical instructions.
Decimal
Decimal arithmetic operates on data in the packed
format. In this format, two decimal digits are placed
in one eight-bit byte. Since data are often communicated
to or from external devices in the zoned format (which has
one digit in an eight-bit byte), the necessary format-
conversion operations are also provided in this instruction
group.
Data are interpreted as integers, right-aligned in
their fields. They are kept in true notation with a sign
in the low-order eight-bit byte.
Processing takes place right to left between main-
storage locations. All decimal arithmetic instructions
use a two-address format. Each address specifies the
leftmost byte of an operand. Associated with this address
is a length field, indicating the number of additional
bytes that the operand extends beyond the first byte.
HARDWARE CHARACTERISTICS
Central Processing Unit
7.E.1.5.
-366-
Approved For Release gpar+5/08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/o?fltk-RDP78-03952A000100070001-5
The decimal arithmetic instruction set provides for
adding, subtracting, comparing, multiplying, and dividing,
as well as the format conversion of variable-length
operands. Most decimal instructions are part of the
decimal feature.
The condition code is set as a result of all add-
type and comparison operations and can be checked by
logical instructions.
Floating-Point Arithmetic
The purpose of the floating-point instruction set
is to perform calculations using operands with a wide
range of magnitude and yielding results scaled to preserve
precision.
A floating-point number consists of a signed exponent
and a signed fraction. The quantity expressed by this
number is the product of the fraction and the number 16
raised t the power of the exponent. The exponent is
expressed in excess 64 binary notation; the fraction is
expressed as a hexadecimal number having a radix point
to the left of the high-order digit.
To avoid unnecessary storing and loading operations
for results and operands, four floating-point registers
are provided. The floating-point instruction set pro-
HARDWARE CHARACTERISTICS
Central Processing Unit
7.E.1.5.
-367-
Approved For Release 2000/05/0810DP78-03952A000100070001-5
Approved For Release 20011*/08 : CIA-RDP78-03952A000100070001-5
vides for loading, adding, subtracting, comparing,
multiplying, dividing, and storing, as well as the sign
control of short or long operands, On the other hand
long operands provide greater accuracy of computation.
Operations may be either register to register or storage
to register. All floating-point instructions and registers
are part of the floating-point feature.
To preserve maximum precision, addition, subtrac-
tion, multiplication, and division are performed with
normalized results. Addition and subtraction may also
be performed with unnormalized results. Normalized and
unnormalized operands may be used in any floating-point
operation.
The condition code is set as a result of all sign
control, add, subtract, and compare operations.
7.E.1.6. Interrupt Structure
An interrupt system alerts the system to an exceptional
condition, the end of an I/O operation, program errors,
machine errors, etc., and sends the system to the appro-
priate routine following the detection of such an event.
The system must have, in effect, the ability to pause to
answer the telephone and then to resume the interrupted
work.
HARDWARE REQUIREMENTS
Central Processing Unit
7.E.1.6.
-368-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/H9FITA-RDP78-03952A000100070001-5
The interruption system makes possible the operation
of a system in a non-stop environment and greatly aids
the efficient use of I/O equipment. The desire to make
the interruption procedure as short and simple as possible
means that the method of switching between the interrupted
program and the program that services the interruption must
be quite efficient. It operates as follows:
The complete status of the System/360 is held in
eight bytes of information. This status information,
which consists of the instruction counter, condition code,
storage protection key, etc., is saved when an interruption
occurs, all the status information, together with an
identification of the cause of the interruption, is
combined into a double (eight-byte) word called the
program status word (PSW).
The PSW is stored at a fixed location, the address
of which depends on the type of interruption. The
system then automatically fetches a new PSW from a
different fixed location, the address of which is also
dependent on the type of interruption. Each class of
interruption has two fixed locations in main storage:
one to receive the old PSW when the interruption occurs,
and the other to supply the new PSW that governs the
servicing of that class of interruption.
HARDWARE REQUIREMENTS
Central Processing Unit
7.E.1.6.
-369-
Approved For Release 2000/05/0etiEWDP78-03952A000100070001-5
Approved For Release 200613VROB : CIA-RDP78-03952A000100070001-5
After the interruption has been serviced, a single
instruction uses the stored PSW to reset the processing
unit to the status it had before the interruption.
The interruption system separates interruptions into
five classes:
Program interruptions are caused by various kinds
of programming errors; the exact type of error is dis-
played in the old PSW.
Supervisor Call interruptions are caused by the
program issuing an instruction to turn over control to
the supervisor (a type of master control program). The
exact reason for the call is shown in the old PSW.
External interruptions are caused by either an
external device requiring attention or by the timer going
to zero. (The external device may be a signal from any
device communicating via the direct control feature.)
Machine Check interruptions are caused by the
machine-checking circuits detecting a machine error.
The system is automatically switched to a diagnostic
procedure.
I/O interruptions are caused by an I/O unit ending
an operation or otherwise needing attention. Identifications
of the device and channel causing the interruption are
HARDWARE REQUIREMENTS
Central Processing Unit
7.E.1.6.
-370-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05K4c161-A-RDP78-03952A000100070001-5
stored in the old PSW; in addition, the status of the
device and channel is stored in a fixed location.
Some of the program interruptions, all external
interruptions, and all I/O interruptions can be masked
(ignored); when this is done, the external and I/O
interruptions are held pending and taken at a later time
(governed by the control program).
During the execution of an instruction, several
interruptive events may occur simultaneously. When
this occurs, the competing interruptions are serviced
in a fixed order of priority.
7.E.2. 2361 CORE STORAGE
This directly addressable core storage may be added
to an IBM System/360. Model 50, 60, 62 or 70. These
units are not input/output units. This low-cost auxiliary
bulk storage may be used as an extension of main storage,
as shared storage for two systems, or as intermediate
storage between main storage and input/output devices.
It facilitates real-time, multiprogrammed, and communi-
cations processing, as well as multiprocessing. This
storage is addressed contiguously with main storage.
Speed of this storage is 8 microseconds. The number of
bytes obtained per storage access, and all other features
HARDWARE REQUIREMENTS
2361 Core Storage
7.E.2.
-371-
Approved For Release 2000/05/0PtafyRDP78-03952A000100070001-5
Approved For Release 2066M-68 : CIA-RDP78-03952A000100070001-5
of large-capacity storage except cycle time, are the
same as those of the main storage of the system.
The model 1 contains 1,048,576 bytes of storage,
and the Model 2 contains 2,097,152 bytes.
A maximum of four Model 2's may be added to the
System/360, Models 60, 62 and 70, to provide 8,388,608
positions of directly addressable core storage. A
maximum of two Model 2's may be added to the System/360
Model 50 to provide 4,194,304 positions of directly
addressable core storage.
Direct addressing of this large storage is accom-
plished by having the effective address of an instruction
determined by either the 24 bit contents of a general
register or the sum of a 12 bit displacement field and
the 24 bit contents of up to two general registers.
7.E.3. I/O CHANNELS
Each System/360 channel directs the flow of informa-
tion between main storage and those input/output devices
the channel controls. The channel relieves the processing
unit of the tasks of communicating directly with I/O and
permits data processing to proceed concurrently with
input/output operations. Each channel has facilities for:
HARDWARE REQUIREMENTS
I/O Channels
7.E.3.
-372-
Approved For Release %.(i.NR/08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/6?%1A-RD1278-03952A000100070001-5
- Accepting I/O instructions from the CPU
- Addressing devices specified by I/O instructions
- Fetching channel control information from
storage
- Decoding control information
- Testing control information for validity
main
Executing control information
- Providing control signals to the I/O interface
- Accepting control-response signals from the
I/O interface
Buffering data transfers
- Checking parity of bytes transferred
- Counting the number of bytes
Accepting status information
transferred
from I/O devices
- Maintaining channel-status information
- Sending, requested status information to main
storage
- Sequencing interruption requests from I/O devices
- ignaling interruptions to the CPU
The System/360 has two types of channels: multi-
plexor and selector. The channel facility necessary to
sustain an operation with an I/O device is called a sub-
channel. The selector channel has one subchannel; the
multiplexor channel has multiple subchannels.
HARDWARE REQUIREMENTS
I/O Channels
7.E.3.
-373-
Approved For Release 2000/05/08 RiikEIRDP78-03952A000100070001-5
Approved For Release 200008838 : CIA-RDP78-03952A000100070001-5
7.E.3.1. Selector Channels
Selector channels are normally used where high-
speed devices (for example, magnetic tape, files and
drums) are attached to a system. Selector channels
transmit bursts of data to or from a single input/output
device at a time. The bursts extend over the whole block
of data or, when command chaining is specified, over the
whole sequence of blocks.
The channel facilities required for sustaining an
input/output operation are termed a "subchannel."
The selector channels have only one subchannel and operate
only one data transmission at a time. Other input/output
devices attached to the channel can in the meantime execute
operations that do not involve communications with the
channel.
7.E.3.2. Multiplexor Channel
The multiplexor channel separates the operations of
high-speed devices from those of lower-speed devices.
Operations on the channel are in two modes: a "multi-
plex" mode for lower data rates, and a "burst" mode for
the higher.
In the multiplex mode, the single data path of the
channel can be time-shared by a large number of low-speed
HARDWARE REQUIREMENTS
I/O Channels
7.E.3.2.
-374--
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/0%E%U-RDP78-03952A000100070001-5
I/O devices operating simultaneously; the channel receives
and sends data to them on demand. When operating in the
burst mode, however, a single I/O device captures the
multiplexor channel and does not relinquish it from the
time it is selected until the last byte is serviced.
Examples of low-speed devices that can operate
simultaneously on the multiplexor channel are:
- Printers
- Card Punches
- Card Readers
- Terminals
Examples of I/O devices that operate in the burst
mode are tape units and disk, drum, or data cell storage.
7.E.4. BULK STORAGE DEVICES AND ASSOCIATED CONTROL UNITS
I/O devices used for bulk system storage include
tape (2400, 7330), disks (1302 and 2311), and data cells
(2321).
7.E.4.1. Tape Units 2401, 2402, 2403, 24)4
The 2400 series tapes are 9-track magnetic tapes,
specially designed for use on the System/360. All tape
units can operate in the read-backward mode. Recording
density is Pno bytes per inch; inter-record gap
HARMARE REQUIREMENTS
Storage Devices/Control Units
7.E.4.1.
-375-
Approved For Release 2000/05/08 :9S4.ERDP78-03952A000100070001-5
Approved For Release 2000/1gRigr: CIA-RDP78-03952A000100070001-5
is .6 inches. Single track errors are corrected auto-
matically on the reread of the record. Each 2401, 2402,
2403, 2404 is offered in three models with the following
performance.
Model 1
Data Rate 22,500 bytes
per sec.
Model 2 Model 3
45,000 bytes 90,000 bytes
per sec. per sec.
2401
Magnetic Tape Unit
A single drive requiring connection to a
control unit.
2402
Magnetic Tape Unit
Two independent drives housed in a single
cabinet requiring connection to a control unit
2403
Magnetic Tape Unit and Control
A single drive which has a self-contained control
unit that allows reading or writing of magnetic
tape on a single channel.
2404
Magnetic Tape Unit and Control
A single drive which contains a control unit that
allows simultaneous reading and writing on two
channels.
Each tape control can control as many as eight tape
drives. The simultaneous control attaches to two channels
in a manner that permits a read-type operation on one tape
drive to be overlapped with a simultaneous write, write
tape mark, or erase tape operation on another drive.
HARDWARE REQUIREMENTS
Storage Devices/Control Units
7.E.4.1.
-376-
Approved For Release 20s0Eini08 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/05/08SWRDP78-03952A000100070001-5
Controls attach to a selector of multiplexor channel and
operate in the burst mode.
Tape controls are also available in two stand-alone
versions:
2803
Tape Control Unit
Allows read or write operations on a single channel
2804
Tape Control Unit
Allows simultaneous reading and writing on two
channels.
Any model of tape drive can be controlled by any
tape control. Characteristics of the three tape drive
models are shown below.
Characteristic Model 1 Model 2 Model 3
Data Rate (kb/kd)* 22.5/45 45/90 90/180
Tape Speed (ips) 28.125 56.25 112.5
Density (bpi) 200 800 ROO
Nominal Inter-record. Gap
(inches)
Nominal Inter-record Cap
.6
.6
.6
Time (millsec)
21.2
10.6
5.3
Rewind Time (minutes)
3.2
2.6
1
Rewind Unload Time (minutes)
2.2
2.2
1
Tape Width (inches).D
r
.9
.5
Number of Tracks
9
9
9
* kb = thousands of bytes per second
kd = thousands of decimal digits per second
ips = inches per second
bpi = bytes per inch
HARDWARE REQUIREMENTS
Storage Devices/Control Units
7.E.4.1.
-377-
Approved For Release 2000/05/08 : suteDP78-03952A0001 00070001 -5
Approved For Release 200 Q$ : CIA-RDP78-03952A000100070001-5
7.E.4.2. 2816 Switching Unit
With the 16-drive control feature installed on the
appropriate tape controls and the addition of two IBM
2816 Switching Units, a maximum of 16 tape drives may
be controlled in common by a maximum of four tape controls.
The minimum is control of four tape drives by two tape
controls; any combination between the minimum and maxi-
mum is available. For more than eight drives, the second
switching unit is required.
Model 1 of the switching unit is used for switching
tape drives in the 2400 series. The drives in the tape-
switching pool may be located in any model of 2401, 2402,
or 2403. The tape controls involved may be in the IBM
2403 Magnetic Tape Unit and Control or the 2803 Tape
Control.
If the 16-drive control feature is not installed on
a tape or Hypertape control, it can control only the
first eight drives in the tape or Hypertape pool.
7.E.4.3. 1302 Disk Storage Unit
The IBM 1302 Disk Storage can record and retrieve
data either at random or sequentially, the random capa-
bility permits immediate access to specific areas of
information without the need to sequentially examine all
HARDWARE REQUIREMENTS
Storage Devices/Control Units
7.E.4.3.
-378--
Approved For Release 2000/05/08 : CIA-RDP78-03952A0Q0100070001-5
SECRET
Approved For Release 2000/05/0855?KiRDP78-03952A000100070001-5
the data in the file. The speed of the access enables
the user to maintain up-to-the-second data files and
make frequent retrievals of the data stored.
The 1302 Model N 1 contains one disk storage
module; the Model N 2 contains two disk storage modules.
Capacity per module is 112.14 million bytes (or 224.28
million pae6ed decimal digits and signs). Eight accesses
(four modules) may be connected to the System/360 through
an IBM 2841 Storage Control Unit, for on-line random
access to 448.56 million bytes (or 897.12 million packed
decimal digits and signs) per 2841 attached. Each module
has two access mechanisms with each access accessing 250
cylinders. A special feature on the control unit permits
eight more access mechanisms, with four additional modules
of disk storage, to be addressed by that control unit.
The recording medium consists of thin, magnetically-
coated metal disks that spin at 1,790 revolutions per
minute. The transfer rate between the 1302 and the
processing unit is 156 thousand bytes per second.
For accessing the 250 cylinders accessed by one access
mechanism are divided into five groups:
1st group has 6 areas of 10 cylinders each.
2nd group has 4 areas of 10 cylinders each.
HARDWARE REQUIREMENTS
Storage Devices/Control Units
7.E.4.3.
-379-
Approved For Release 2000/05/08 : sm.9P78-03952A000100070001-5
Approved For Release 20003135N8 : CIA-RDP78-03952A000100070001-5
3rd group has 6 areas of 10 cylinders each.
4th group has 4 areas of 10 cylinders each.
5th group has 5 areas of 10 cylinders each.
To access a track in:
- Same cylinder is 0 ms
- Another cylinder in same 10 cylinder area is 50 ms.
- Another area in same group is 120 ms.
- Another group is 180 ms.
This gives an average access time of 165 ms.
7.E.4.4. 2311 Disk Storage Drive
The IBM 2311 Disk Storage Drive provides random access
storage for 7.25 million bytes (or 14.5 million packed
decimal digits and signs) on a single disk pack. Eight
disk storage drives can be attached to each IBM 2841
Storage Control Unit, for a total on-line capacity of
58 million bytes per 2841. In addition, unlimited storage
capacity is possible because the disk pack in each drive
can be easily removed and replaced with another in less
than I minute.
The disk pack weighs only 10 pounds; each has six
14-inch disks, mounted 1/2 inch apart on a vertical
shaft. The inside ten disk surfaces are used for
recording data and the outermost two sufaces are pro-
HARDWARE REQUIREMENTS
Storage DeVices/Control Units
7.E.4.4.
-380-
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5
SECRET
Approved For Release 2000/05/08sFERTRDP78-03952A000100070001-5
tective plates. When the disk pack is installed in the
disk storage, information is written on or read from
the disk sufaces by magnetic read/write heads, mounted in
pairs between each two disks on a movable comb-like access
mechanism.
The data rate of the IBM 2311 is 156 thousand bytes
per second. Sequential track-to-track access time is
30 milliseconds. The maximum seek time is 145 milli-
seconds, and the average seek time is 85 milliseconds.
The average rotational delay is 12.5 milliseconds.
7.E.4.5. 2321 Data Cell Drive
Each 2321 Data Cell Drive can hold 400 million
bytes (or 800 million packed decimal digits and signs)
of on-line data. Eight 2321's may be attached to a
2841 Storage Control Unit, for a total on-line capacity
of 3.2 billion bytes or 6.4 million pack decimal digits
and signs per 2841 control unit. The data rate is
55 thousand bytes per second.
Each data cell drive contains from one to ten data
cells, each having a capacity of 40 million bytes. The
data cells are all removable and interchangeable,
permitting an open ended capacity for libraries of data
cells.
HARDWARE REQUIREMENTS
Storage Devices/Control Units
7.E.4.5.
-381-
Approved For Release 2000/05/08 :yeepP78-03952A000100070001-5
Approved For Release 200
I.
CIA-RDP78-03952A000100070001-5
CONFIDENTIAL
The storage medium is a strip of magnetic tape
2 1/4 inches wide by 13 inches long. Each data cell
contains 200 of these strips, divided into 20 subcells
of 10 strips each. A rotary positioning system aligns
the selected subcell beneath the access station. As
shown below, access time to a strip varies from 175
milliseconds to 600 milliseconds, depending on the
number of subcells the array must move.
Number of Subcells Array Must
Move
0 1 50 100 (max.)
No strip on drum 175 ms 250 ms 350 ms 400 ms
Previously addressed 375 ms 450 ms 550 ms 600 ms
strip on drum
7.E.4.6. 2841 Storage Control
The IBM 2841 Storage Control Unit provides the
capacility of attaching serial-by-bit types of random access
storage devices to the IBM System/360. Adapters are
available for the 2841 to control these storage units:
- IBM 1302 Disk Storage Models Ni, N2
- IBM 2311 Disk Storage Drive
- IBM 2321 Data Cell Drive
Approved For Release 20
HARDWARE REQUIREMENTS
Storage Devices/Control Units
7.E.4.6.
-382- CONFIDENTIAL
8 : CIA-RDP78-03952A000100070001-5
Approved For Release 2000/0451VIAETA-RDFCMENTiAr70001-5
Adapters can be mixed to provide 2841 control of up to
eight access mechanisms, in any combination, on the above
units.
ACCESS MECHANISMS PER FILE
1302 Ni Two Access Mechanisms (one module)
1302 N2 Four Access Mechanisms (two modules)
2311 One Access Mechanism
2321 One Access Mechanism
FILES PER 2841
(One of the following, when adapters are not intermixed)
1302 Ni 4
1302 N2 2
2311 8
2321 8
The 2841 interprets and executes all orders from the
system, and checks the validity of the data transferred to
and from the storage devices.
7.E.5. SYSTEM COMMUNICATION DEVICES*'
Various IO devices are needed for convenient communica-
tion between the user and computer system. Some of these
IO devices available on the 360 are discussed in this
section.
HARDWARE REQUIREMENTS
Communication Devices
7.E.5.
-383-
Approved ForRelease2000/05/0W-RDP78-03952A000100070001-5
CONFinciuri II
AP Poof ty
2000/6 IA-RDP78-03952A000100070001-5
7.E.5.1. 1442 Card Read Punch
The IBM 1442 Card Read Punch Model Ni uses a single
common card path for reading and punching, and reads and
punches cards serially by card column. Serial card
feeding, past a read and then a punch station, makes it
possible for the program to read data from a card, then
hold the card at the next consecutive card column, and
then punch data (such as the results of a calculation)
into the same card, all during a single card pass.
The IBM 1442 Model Ni reads or punches cards at the
following maximum rates:
Reading 400 cards per minute
Punching 160 card columns per second
The 1442 is an I/O device typically operated on the
multiplexor channel. The control circuits are within
the 1442.
The 1442 Model N1 reads and punches an extended
card code that permits punching any of 256 different
combinations of holes in each card column. The optional
card image feature permits the low-order six bits of
bytes read from the processing unit to be punched into
the upper six or lower six rows of a card, enabling
160 bytes to be placed in the card. On reading, the
information is read column by column and transmitted to
-384-
/08 : ttliFTVEC3V*00100070001.5
Approved For Release 2000
HARDWARE REQUIREMENTS
Communication Devices
7.E.5.1.
Approved For Release 2000/05/W16X-RDP78-43g52aQQ0100070001-5
WINI-IDENTIAL
the processing unit byte by byte, with two high-order
bits (6 and 7) being set to zero. Interspersed cards
punched in either the normal or the card image mode may
be read.
The read hopper holds 1,200 cards. Card movement
is from the hopper to one of the two stackers (pockets).
The stackers hold 1,200 cards each, and cards can be
removed from either stacker without stopping the machine.
7.E.5.2. 2201 Printer
The IBM 2201 Printer Model 3 printer has 132 print
positions and a speed of 1,100 lines per minute.
In the standard character set, there is a regular
distribution of the 48 characters in five identical groups.
In the optional preferred character set, the 48 characters
are distributed irregularly, with more of some and fewer
of others than in the standard character set. This
produces print rates of three levels, depending on the
characters printed:
1400 lines per minute for the 10 numeric digits
and .,*-
950 lines per minute for the 26 letters and
500 lines per minute for the characters + @ # +
Approved For Release 2000/05/0
-385--
HARDWARE REQUIREMENTS
Communication Devices
7.E.5.2.
RDP78-03
9c212
ApproOLEIWiftt?
IA-RDP78-03952A000100070001-5
When characters from different levels are inter-mixed,
the speed becomes variable at rates between the levels.
The 2201 is controlled, buffered, and attached to
the System/360 by an IBM 2821 Control Unit Model 1 or 2.
One or two additional 2201 printers can be serviced by a
2821 Model 3, with a 2821 Model 1 or 2 as a prerequisite.
7.E.5.3. 1052 Console Typewriter
The IBM 1052 Printer-Keyboard Model 1, part of the
1050 Data Communication System, can be attached to the
system for communication between remote locations and the
system, and the operator and the system.
The keyboard and the printing are electrically and
mechanically independent of each other. The keyboard
can be used for system input, and the printer accepts
computer output. A single keyboard depression causes
an I/O interruption, and a double keyboard depression
signals the end of data entry.
The 1052 prints at 14.8 characters per second.
It has replaceable printing head and a typewriter style
keyboard.
7.E.5.4. 1050 Data Communication System
The IBM 1510 DaEa Communication System is a multi-
purpose office-oriented TELE-PROCESSING system. It can
-386-
Approved For Release 2000/15/08 : CIAIRDP78=03952A000100070001-5
,4.
HARDWARE REQUIREMENTS
Communication Devices
7.E.5.4.
CONFIDENTIAL
Approved For Release 2000/05/0WKR
/41414)yudl'4'070001-5
be operated over leased communications lines (through
the IBM 2701 Data Adapter Unit or 2702 Transmission
Control) to connect the System/360 to remote locations.
The versatility of the 1050 system is furthered by
its variety of configurations. The choice of configura-
tions ranges from a receiver-printer to a configuration
that includes:
- Manual keyboard entry
- Printed page
output
- Punched-card input and output
- Paper-tape input and output
- Edge-punched-document input and
output
The component type numbers and names are:
- 1051 Control Unit
- 1052 Printer-Keyboard
- 1053 Printer
- 1054 Paper Tape Punch
- 1055 Paper Tape Punch
- 1056 Card Reader
- 1057 Card Punch
- 1058 Printing Card Punch
In addition to providing on-line transmission at
14.8 characters per second, the 1050 can also simultaneously
Approved For Release 2000/05/08
HARDWARE REQUIREMENTS
Communication Devices
7.E.5.4.
-387-
rp78-0a0w00100970001 -5
Approved For Release 200
: b014.610E-Khih
,000100070001-5
prepare and record off-line data. Transmission is checked
for accuracy.
The IBM 1051 Control Unit is required in all configura-
tions and contains the power supply, code translator, data
channels, and control circuitry for the 1050 system.
7.E.5.5. 1015 Inquiry Display Terminal
This device and its IBM 1016 Control Unit make it
possible for an operator to see records in main or
auxiliary storage as needed, such as upon customer
demand. The information is written at a rate of up to
650 characters per second, about 40 times faster than
that produced for an operator by means of typeout.
Once written, the information is retained for a long as
desired.
The viewing area of this cathode ray tube has the
capacity for 30 lines of 40 characters each, for a total
of up to 1,200 characters at a time. Horizontal spacing
is ten characters to the inch; vertical spacing, eight
lines per inch. The characters standard for display are:
all 36 alphameric characters plus 23 special characters.
Plotting is available optionally. Plotting resolu-
tion is approixmately 20 points per inch horizontally
and 15 vertically. The plotting matrix is 60 x 80.
HARDWARE REQUIREMENTS
Communication Devices
7.E.5.5.
-
Approved For Release 2s0;34858/08 ? CIA-PIR9PJEHMCIgUA0010100070001-5
Approved For Release 2000/05/1744eRDIfeakalkOWC16670001-5
Four special characters are used for plotting. A line
advance suppress feature is also incorporated into the
plotting option to allow the display of more than one
plotting point in a given character space.
Up to ten 101 Model l's can be attached to the
1016 Control Unit. Each Model 1 allows the attachment
of five 1015 Model 2's providing a maximum of sixty
1015 Inquiry Display Terminals (ten 1015 Model l's and
fifty 1015 Model 2's).
HARDWARE REQUIREMENTS
Communication Devices
7.E.5.5.
-389-
Approved For Release 2000/05/08? RDP78-04QN
ik v ApTpttl -5
Approved For Release 2000/05/08 : C -R
Mlifil3ENT IMP 7 0 0 01-5
:SECRET CONFIDENTIAL
Approved For Release 2000/05/08 : CIA-RDP78-03952A000100070001-5