PDP-10 Archive: 43,50212/ffm.opr from decuslib10-01

Trailing-Edge - PDP-10 Archives - decuslib10-01 - 43,50212/ffm.opr

There are no other files named ffm.opr in the archive.

A FORTRAN FILE MAINTENANCE SYSTEM

MYRON W. CURTIS
COMPUTING CENTER
BOWDOIN COLLEGE
BRUNSWICK MAINE 04011

INTRODUCTION

THE SYSTEM IS DESIGNED FOR A COMPUTER WITH A RANDOM ACCESS STORAGE DEVICE
(DISK OR CORE). THE SYSTEM CONSISTS OF A SET OF CONVENTIONS FOR FILE DESIGN AND
A SET OF FORTRAN SUBROUTINES THAT OPERATE ON AN ARBITRARY FILE. THESE
SUBROUTINES DO MOST OF THE ROUTINE JOBS NECESSARY TO KEEP DATA FILES UP TO DATE
AND TO WRITE REPORTS FROM THESE FILES. THE SUBROUTINES ARE WRITTEN IN FORTRAN
WITH THE EXCEPTION OF THE DISK INPUT/OUTPUT ROUTINE WHICH IS A MACHINE LANGUAGE
ROUTINE. OF COURSE, ANY OF THE SUBROUTINES MAY BE REWRITTEN IN MACHINE LANGUAGE
AT A PARTICULAR INSTALLATION IN ORDER TO CONSERVE EXECUTION TIME.
THE SYSTEM WAS WRITTEN AT BOWDOIN COLLEGE IN FORTRAN II FOR THE I.B.M.
1620 AND HAS BEEN MODIFIED AND EXTENDED IN FORTAN IV FOR THE D.E.C. PDP-10

FILE CONVENTIONS

THE FILE MAINTENANCE SYSTEM IS COMPLETELY DEPENDENT ON A SET OF CONVENTIONS
FOR DESIGNING FILES WITHIN THE SYSTEM SINCE THE MAINTENANCE SUBROUTINES ARE
PROGRAMMED WITH THESE CONVENTIONS IMPLICITLY IMBEDDED IN THE CODE. SO TO MODIFY
A CONVENTION WOULD IN GENERAL REQUIRE A CORRESPONDING MODIFICATION IN THE SYSTEM
PROGRAMS.

**DEFINITION 1** - A W O R D IS A FORTRAN VARIABLE AND HENCE, MAY BE IN THREE
MODES -- (1) INTEGER, (2) FLOATING POINT, AND (3) ALPHANUMERIC

**CONVENTION 1** - ALL WORDS MUST HAVE THE SAME LENGTH. IN FIXED WORD LENGTH
COMPUTERS, THIS IS INHERENT IN THE NATURE OF THE COMPUTER. IN THE 1620,
ONE COMPILES ALL PROGRAMS WITH INTEGER VARIABLES OF 10 DIGITS AND FLOATING
POINT VARIABLES OF 8 DIGIT FRACTIONS AND 2 DIGIT EXPONENTS. (*FANDK0810)

**DEFINITION 2** - A F I E L D (SOMETIMES REFERRED TO AS C A T E G O R Y) IS
A CONSECUTIVE STRING OF WORDS IN THE SAME MODE. THEREFORE, WE MAY DESCRIBE
A FIELD BY THREE WORDS -- (1) THE LOCATION OF THE FIRST WORD, (2) THE
LENGTH OF THE FIELD OR NUMBER OF WORDS IN A FIELD, AND (3) THE MODE OF THE
FIELD.

**DEFINITION 3** - A R E C O R D IS A CONSECUTIVE STRING OF F I E L D S. IT
IS SOMETIMES CONVENIENT TO THINK OF A RECORD AS A CONSECUTIVE STRING OF
WORDS. IN THIS SYSTEM, A RECORD IS REFERRED TO BY THE ONE-DIMENSIONAL
ARRAY I R. HENCE, THE ITH WORD IS I R(I).
ONE SHOULD NOTE THAT CONSECUTIVE, IN ADDITION TO ITS USUAL DEFINITION,
MEANS STORED CONSECUTIVELY IN MEMORY OR ON THE DISK.

**DEFINITION 4** - A F I L E IS A CONSECUTIVE STRING OF R E C O R D S. EACH
RECORD IN A FILE MUST HAVE THE SAME NUMBER OF FIELDS OF CORRESPONDING MODE
AND ORDER. ONE MIGHT SAY THEY ARE CONGRUENT.

**CONVENTION 2** - THE FIRST FIELD IN EACH RECORD IS ASSUMED TO BE AN
IDENTIFICATION FIELD CONSISTING OF ONE WORD. THERE IS NO UNIVERSAL
RESTRICTION ON MODE, BUT OF COURSE THE I D OF EACH RECORD OF A FILE MUST BE
OF THE SAME MODE. SOME UTILITY ROUTINES ASSUME THE ID IS IN INTEGER MODE.
**DEFINITION 5** - A P A C K IS A CONSECUTIVE STRING OF F I L E S. USUALLY
A PACK IS A CONTIGUOUS STORAGE AREA ON A DISK PACK (PACKS) OR IN A LARGE
MEMORY. WE WILL ASSUME FROM NOW ON THAT A PACK IS LOCATED ON A DISK.

**DEFINITION 6** - A S E C T O R IS THE AMOUNT OF DISK STORAGE NECESSARY TO
STORE TEN CONSECUTIVE WORDS AND MUST BE LOGICALLY ADDRESSABLE BY THE
COMPUTER IN QUESTION.

**CONVENTION 3** - ALL RECORDS, FILES, AND PACKS TAKE UP AN INTEGRAL NUMBER
OF SECTORS ON THE DISK. THIS MEANS THAT ALTHOUGH THE NUMBER OF WORDS IN A
RECORD MAY BE LESS THAN 10K, THE RECORD TAKES UP SPACE FOR 10K WORDS ON THE
DISK.
DATA ON THE DISK IS LOCATED BY SECTOR NUMBER FROM A FIXED ADDRESS ON
THE DISK CALLED THE O R I G I N. THIS ADDRESS IS DETERMINED BY DIO (A
MACHINE LANGUAGE DISK I/O ROUTINE).
LOCATION (ADDRESS) OF RECORDS AND FILES IN THE SYSTEM ARE GIVEN BY THE
APPROPRIATE SECTOR NUMBER RELATIVE TO THE ORIGIN.

WE SEE THAT ONE MAY DESCRIBE AN ARBITRARY FILE IN A PACK BY GIVING IT AN
IDENTIFYING NAME, ITS RECORD FORMAT (THE NUMBER AND MODE OF FIELDS), AND THE
LOCATION AND SIZE OF THE FILE. THE FOLLOWING CONVENTIONS DEFINE THESE FILE AND
RECORD PARAMETERS EXPLICITLY.

**CONVENTION 4** - THE RECORD FORMAT SPECIFICATIONS (**DEFINITION 3**) ARE
STORED IN MEMORY IN A 2-DIMENSIONAL ARRAY, CALLED I F R M A T. WHERE
I F R M A T (1,I), I F R M A T (2,I), I F R M A T (3,I) ARE INTEGERS
DEFINING THE LOCATION, LENGTH, AND MODE OF THE ITH FIELD OF A RECORD.
THE LOCATION OF A FIELD IS GIVEN BY A WORD NUMBER RELATIVE TO
THE FIRST WORD OF A RECORD.
I F R M A T (3,I) = (1) - INTEGER
(2) - FLOATING POINT
(3) - ALPHA

**CONVENTION 5** - THIS CONVENTION GIVES THE DEFINITION AND A SUGGESTED
FORTRAN MNEMONIC FOR THE FILE PARAMETERS.
I P A R (1) = I D F - A 5-CHARACTER ALPHA IDENTIFICATION FOR THE FILE.
I P A R (2) = L F R - THE BEGINNING SECTOR NUMBER FOR THE FIRST RECORD
OF THE FILE.
I P A R (3) = N A V R - THE SECTOR NUMBER FOR THE NEXT AVAILABLE
RECORD IN THE FILE.
I P A R (4) = M A X R - THE MAXIMUM SECTOR NUMBER AVAILABLE FOR THE
FILE.
I P A R (5) = N S P R - THE NUMBER OF SECTORS PER RECORD.
I P A R (6) = L S R - THE SECTOR NUMBER, OR ADDRESS, ONE MIGHT SAY THE
SCANNED RECORD ADDRESS, OF A RECORD IN THE FILE
TO BE READ OR WRITTEN.
I P A R (7) = L F M T - THE SECTOR NUMBER ADDRESSING THE RECORD FORMAT
SPECIFICATION.
I P A R (8) = N C P R - THE NUMBER OF FIELDS (CATEGORIES) PER RECORD.
I P A R (9) = L P F R - THE LOCATION OF THE POINTER FILE RECORD WHERE THE
FILE PARAMETERS ARE STORED
I P A R (10) = DATME - DATE AND TIME OF LAST "CALL SAVEF"

**CONVENTION 6** - THE FIRST FILE IN A PACK IS CALLED THE P O I N T E R
F I L E. ITS RECORDS CONSIST OF THE 10 FILE PARAMETER WORDS IN THE ABOVE
CONVENTION. THE FIRST RECORD OF THE POINTER FILE IS THE RECORD OF FILE
PARAMETERS DESCRIBING THE POINTER FILE. THE I D OF THE FIRST RECORD IS
CONSIDERED THE I D FOR THE PACK. THE LAST RECORD OF THE POINTER FILE
DESCRIBES THE MASTER FILE AND THE MASTER FILE PARAMETERS DESCRIBE THE
COMPLETE SET OF DATA ON A PACK ONE SECTOR PER RECORD.
THE CONCEPT OF THE MASTER FILE IS USED TO ENABLE ONE TO MANIPULATE FILES ON
THE PACK. M A X R OF THE MASTER FILE IS THE MAXIMUM SECTOR ADDRESS
AVAILABLE ON A PACK.
IF WE ALLOW 20 FILES ON OUR PACK AND 20,000 SECTORS, INITIALLY THE POINTER FILE IS GIVEN
BY THE 20 X 10 MATRIX
PACK1 1 3 20 1 1 0 0 1 0
0 21 21 20000 1 21 0 0 2 0
. .
. .
. .
0 . . . .
WHERE THE ITH ROW CORRESPONDS TO THE FILE PARAMETERS OF THE ITH FILE ON THE
PACK. SINCE THE RECORD FORMAT FOR THE POINTER FILE AND THE MASTER FILE IS ALWAYS THE SAME, IT
IS NOT STORED ON THE DISK, AND NCPR=0.

WE SEE THAT A FILE IN THE SYSTEM MAY BE VIEWED AS A 2-DIMENSIONAL MATRIX,
F I L E, WHERE F I L E (I,J) IS THE JTH WORD OF THE ITH RECORD. IN GENERAL,
THERE IS NOT ENOUGH ROOM IN CORE TO STORE AN ENTIRE FILE SO THAT ONE RECORD AT
A TIME IS USUALLY READ INTO MEMORY INTO A 1-DIMENSIONAL ARRAY, I R, WHERE
I R (J) IS THE JTH WORD OF THE RECORD.

**CONVENTION 7** - THE ADDRESS (SECTOR NUMBER) OF THE FIRST RECORD OF THE
FILE MINUS THE NUMBER OF SECTORS PER RECORD IS THE ADDRESS OF A SCRATCH
RECORD. IN FORTRAN, THIS ADDRESS MIGHT BE GIVEN BY I A D R E S = L F R -
N S P R.

PROGRAMMING CONVENTIONS

THE GENERAL PROGRAMMING PHILOSOPHY IN THE FILE MAINTENANCE SYSTEM IS THAT
THE FILE AND RECORD PARAMETERS DESCRIBING THE FILE BEING OPERATED ON BE IN CORE.
THESE ARE STORED IN C O M M O N.

**CONVENTION 8** - EACH PROGRAM IN THE SYSTEM MUST DEFINE C O M M O N STORAGE
TO ACCOMMODATE THE ABOVE FILE AND RECORD PARAMETERS. HENCE, THE FOLLOWING
FORTRAN STATEMENTS SHOULD PRECEDE ANY OTHER C O M M O N STATEMENTS IN A
MAIN PROGRAM.

C O M M O N IDF,LFR,NAVR,MAXR,NSPR,LSR,LFMT,NCPR,LPFR,DATME
C O M M O N IFRMAT(3,N)
E Q U I V A L E N C E (IDF,IPAR(1))

(WHERE N IS THE LARGEST NUMBER OF FIELDS PER RECORD IN ANY FILE
BEING USED). NOTE IF THE PROGRAM IS A SUBROUTINE, I F R M A T
MAY BE ONE-DIMENSIONAL AND THE SUBROUTINE I S U B (I,J) GIVES
THE CORRESPONDING ONE-DIMENSIONAL SUBCSRIPT.

THE USER SHOULD NOTE THAT THERE ARE TWO WAYS OF REFERRING TO THE FILE
PARAMETERS IN A FORTRAN PROGRAM. FOR EXAMPLE, TO COMPUTE THE NUMBER OF
RECORDS IN A FILE ONE MIGHT WRITE
N R = (N A V R - L F R) / N S P R
OR
N R = (I P A R (3) - I P A R (2) ) / I P A R (5)
TO ALLOW FOR FIELDS OF BOTH INTEGER AND FLOATING POINT MODE IN THE SAME
RECORD. WE DEFINE A BUFFER REGION (A 1-DIMENSIONAL ARRAY) AND GIVE IT BOTH
INTEGER AND FLOATING POINT MODE. IF I R IS THE NAME OF THE RECORD BUFFER
WE HAVE

**CONVENTION 9** -
D I M E N S I O N I R (N), R (N) ( WHEREN IS EQUAL TO 10*NSPR

E Q U I V A L E N C E (I R(1), R(1))

NOTE THAT IF A SUBROUTINE NEEDS BOTH INTEGER AND FLOATING FIELDS IN A
RECORD, THEN BOTH I R, AND R MUST APPEAR IN THE CALLING SEQUENCE.

THE BULK OF THE SYSTEM CONSISTS OF A SET OF FORTRAN SUBROUTINES AND
FUNCTIONS WRITTEN IN EITHER FORTRAN OR ASSEMBLY LANGUAGE THAT
PERFORM THE BULK OF THE WORK IN ANY OPERATION ON A DATA FILE. THESE PROGRAMS
COME IN THE FOLLOWING CLASSES -
(1) FILE DEFINITION AND MANIPULATION
(2) RECORD DEFINITION AND MANIPULATION
(3) RECORD SEARCHING
(4) WORD MANIPULATION
(5) GENERAL UTILITY AND DEBUGGING PROGRAMS.

THE USER SHOULD NOTE THE LIBRARY OF THESE PROGRAMS SHOULD IN NO WAY BE
CONSIDERED FIXED. PROGRAMS MAY BE ADDED AND IMPROVED AT ANY TIME AS LONG AS
THEIR APPEARANCE TO THE USER REMAINS THE SAME.

THE FOLLOWING IS A LIST OF THE MOST COMMONLY USED PROGRAMS OF THE SYSTEM.
TYPE DESCRIPTION NAME AND CALLING SEQUENCE
(FUNCTION,
SUBROUTINE,
OR MAIN
PROGRAM)

S DEFINE A FILE DFINEF(IDFILE,NSPR1,MAXR1,NCPR1)
S DEFINE A RECORD DFINER
S SELECT A FILE SLECTF(IDFILE)
S WRITE A RECORD ON DISK WRITER(IR)
S READ A RECORD ON DISK READR(IR)
S SAVE FILE PARAMETERS IN
POINTER FILE SAVEF
S CLEAR RECORD BUFFER TO ZERO CLEAR(IR)
S ADVANCE N RECORDS IN FILE RIGHT(N)
S RETREAT N RECORDS IN FILE LEFT(N)
F FIND RECORD ON DISK IFIND(IDR,IDIR,IHOW)
S CLEAR A FILE TO ZEROS CLEARF(IDFILE)
S TO DELETE A RECORD FROM FILE DELETR(IDR,IR)
S TO INSERT A RECORD IN A FILE INSRTR(IR,IHOW)
F SELECTIVE SEARCH IGETR(IR,N,INDEX,IVALUE,IF,LOGIC)
S TO DUMP A RECORD WORD BY WORD DUMP(IR)
MP TO DELETE A FILE SCRTCH
MP TO EXPAND A FILE INLRGE
S TO DUMP A RECORD CATEGORY BY
CATEGORY DUMPR(IR,R)
MP SELECTIVE TYPING TYPEF
MP SELECTIVE RECORD DUMP DUMPF
MP DEFINE A PACK DPACK
MP DEFINE INITIAL POINTER FILE LODPTF
MP DUMP POINTER FILE PARAMETERS DMPPTF
MP LOAD FORMAT SPECIFICATIONS LODFMT
MP DUMP FORMAT SPECIFICATIONS DMPFMT
MP EDIT CATEGORIES IN A RECORD EDITF
MP DEFINE INITIAL FILE PARAMETERS CREATF
MP SELECTIVE LISTING LISTF

IN NAMING PROGRAMS, THOSE DEALING WITH RECORDS USUALLY END IN R AND THOSE
WITH FILES END IN F.

NOTE THAT VARIABLES DEFINED IN C O M M O N MAY NOT BE USED IN CALLING
SEQUENCES.

THE FOLLOWING IS A DESCRIPTION OF SOME OF THE MOST COMMONLY USED
SUBROUTINES IN THE SYSTEM. FORTRAN VARIABLE NAMES ARE INCLOSED IN **.
QUOTES APPEAR AS ".
TYPE I FILE DEFINITION AND MANIPULATION

1. DFINEF (IDFILE, NSPR1, MAXR1, NCPR1)
THIS PROGRAM SETS UP FILE PARAMETER RECORD IN POINTER FILE AND LEAVES
THEM IN MEMORY UPON EXIT. THE PARAMETERS IN THE CALLING SEQUENCE MUST NOT BE IN
**IPAR**(COMMON).

2. SLECTF (IDFILE)
LOOKS UP **IDFILE** IN POINTER FILE AND READS FILE AND RECORD PARAMETERS
FROM DISK INTO COMMON VARIABLES **IPAR** AND **IFRMAT**.

3) SAVEF
PUTS **IPAR** AND **IFRMAT** BACK ON THE DISK. SHOULD BE USED WHENEVER
**NAVR** OR **LSR** HAVE BEEN CHANGED BY THE PROGRAM AND THE USER IS CALLING IN
A NEW FILE OR THE PROGRAM HAS FINISHED. **IPAR** AND **IFRMAT** REMAIN IN CORE.
THE USE OF THIS PROGRAM ALSO INSURES THAT ANY UPDATED INFORMATION IN
I/O BUFFERS (IN DIO) IS WRITTEN ON THE DISK.

4. CLEARF(IR)
THIS PROGRAM INITIALIZES THE ENTIRE FILE AREA ON THE DISK TO ZEROS IE
IT WRITES A RECORD FILE FULL OF ZEROS FROM **LFR** TO **NAVR**. IR(1) OF
EACH RECORD IS UNCHANGED.

5. MERGEF(IDM,IDT,IR)
**IDM**, **IDT** ARE FILE "ID'S" OF MASTER AND TEMPORARY FILES
RESPECTIVELY. BOTH FILES MUST BE IN ASSENDING ORDER BY THEIR RECORD "ID'S"
THE PROGRAM MERGES THE TEMPORARY FILE TO THE MASTER FILE SO AS TO PRESERVE
**ID** ORDER. THE TEMPORARY FILE IS LEFT UNCHANGED.

6. PULLF (IDM,IDT,IR)
**IDT** IS A FILE WHOSE RECORD "ID'S" CORRESPOND TO SPECIFIC RECORD
"ID'S" OF THE MASTER FILE **IDM**. THE ORDER OF THE RECORD **ID'S** OF
**IDT** MUST BE THAT OF THEIR ORDER OF OCCURANCE IN THE MASTER FILE. THIS PRO-
GRAM EXTRACTS THE SPECIFIED RECORDS FROM THE MASTER FILE AND STORES THEM IN THE
TEMPORARY FILE. THESE RECORDS ARE DELETED FROM THE MASTER FILE.

7. EXPNDF (IDM,N)
THIS PROGRAM EXPANDS THE MAXIMUM CAPACITY OF THE FILE FILE **IDM** BY **N**
RECORDS.

8. DELETF (IDM)
THIS PROGRAM DELETS THE FILE **IDM** FROM THE PACK. WHEN AT ALL POSSIBLE
FILES SHOULD BE DELETED BY THE "LAST IN - FIRST OUT" ALGORITHM.

9. RLSORT(NPACKS,PACK1,ORIGF,PACK2,TEMPF,SORTDF,NWORD,NPASS,NCHAR,
NSEC,MODE)
THIS PROGRAM SORTS A FILE ON SELECTED LOGICAL CHARACTERS OF
A SPECIFIED WORD IN A RECORD. A SECOND PACK ON A DIFERENT
FILE STRUCTURE(LEVEL D) MAY BE USED TO HOLD A TEMPORARY SCRATCH FILE TO
SPEED UP THE SORT. THE SORTING ALGORITHM PRESERVES PREVIOUS
ORDERINGS.
NPACKS-NUMBER OF PACKS USED. 1 OR 2
PACK1- ID OF POINTER FILE OF PACK CONTAINING ORIGF
ORIGF- ID OF MASTER FILE.WILL BE ALTERED BY SORT.
PACK2- ID OF POINTER FILE OF PACK TO HOLD TEMPF
TEMPF- ID OF TEMPORARY FILE USED BY ALGORITHM
SORTDF- ID OF FILE TO PUT SORTED FILE. MUST BE ORIGF,OR TEMPF.
NWORD- WORD NUMBER WITHIN A SECTOR OF RECORD TO SORT ON.
NPASS- NUMBER OF LOGICAL CHARACTERS IN WORD TO SORT ON.
NCHAR- LOGICAL CHARACTER NUMBER FROM RIGHT TO START WITH.
NSEC- SECTOR NUMBER WITHIN RECORD HOLDING KEY WORD.
MODE- TYPE OF LOGICAL CHARACTER TO SORT.
1- 0NE DECIMAL DIGIT
2- TWO DECIMAL DIGITS
3- ONE ASCII CHARACTER

TYPE II RECORD DEFINITION AND MANIPULATION

1. DFINER
THIS SUBPROGRAM WRITES THE CONTENTS OF **IFRMAT** ON THE DISK ACCORDING
TO THE SPECIFICATIONS OF **LFMT** AND **NCPR** OF THE FILE **IDF** WHOSE
PARAMETERS ARE CURRENTLY IN COMMON.

2. CLEAR(IR)
THIS PROGRAM CLEARS THE RECORD STORAGE AREA SPECIFIED BY **IR** TO ZEROS
FOR NUMBERIC WORDS AND BLANKS FOR ALPHABETIC WORDS.

3. READR(IR), WRITER(IR)
THESE PROGRAMS READ (WRITE) THE RECORD FROM (TO) THE DISK SPECIFIED BY
**LSR** (THE LOCATION OF THE SCANNED RECORD). **LSR** MUST BE WITHIN THE
RANGE OF DEFINITION OF THE FILE. **LFR** IS LESS THAN OR EQUAL TO **LSR** IS
LESS THAN OR EQUAL TO **MAXR**

4. INSRTR(IR,IHOW)
THIS SUBPROGRAM INSERTS THE RECORD STORED IN **IR** IN THE FILE ACCORDING
TO **IHOW**.
A) **IHOW=+1. THE PROGRAM SEARCHES THE FILE FROM **LSR** TO THE END AND
AND PLACES THE RECORD IN FRONT OF THE FIRST RECORD IT MEETS WITH AND **ID**
GREATER THAN **IR(1). IF NONE EXIST THE RECORD IS PLACES AT THE END OF THE FILE
**NAVR**.
B) **IHOW**=-1 SIMILIAR TO CASE (A) EXCEPT SEARCH IS MADE TO FIND
THE FIRST RECORD WHOSE **ID** IS LESS THAN **IR(1)**.
C) **IHOW=0. RECORD IS PLACED AT THE END OF THE FILE.

5. DELETR(IDR)
THIS FILE IS SEARCHED FOR A RECORD WHOSE "ID" (IR(1)) IS EQUAL TO
**IDR** AND THAT RECORD IS DELETED FROM THE FILE.
THIS PROGRAM SHOULD ONLY BE USED SPARINGLY ON LARGE FILES.
IF MANY RECORDS ARE TO BE DELETED , USE PULLF.

TYPE III RECORD SEARCHING

1. RIGHT (N), LEFT (N)
ADVANCE, N (BACK SPACE) RECORDS IN THE FILE. NO INFORMATION IS
TRANSMITTED TO OR FROM THE DISK.

2. IFIND(IDR, IDIR, IHOW)
THIS FUNCTION SEARCHES A FILE FOR A RECORD WHOSE ID IS EITHER LESS THAN,EQUAL TO
OR GREATER THAN **IDR** ACCORDING TO DIRECTIONS GIVEN WITH PARAMETERS **IDIR**,
**IHOW**.

**IDIR**=1 MEANS SEARCH IS TO THE RIGHT OR (DOWN A FILE) AND PROCEEDS
RECORD BY RECORD. **IDR**=-1 MEANS SEARCH IS TO THE LEFT. AFTER 15 REOCRDS A
BINARY SEARCH IS CALLED. ***IDR***=0 MEANS SEARCH IS LINEAR TO THE RIGHT
(NO BINARY SEARCH) AND THE FILE IS VIEWED AS A CIRCLE.

**IHOW**=0 THE PROGRAM SEARCHES LINEARLY UNTIL A RECORD IS FOUND WITH
**IR(1)=IDR**. LSR IS SET AT THE APPROPRIATE RECORD. THE FUNCTION RETURNS WITH
VALUE 1. IF RECORD IS NOT FOUND THE FUNCTION EXITS WITH THE VALUE 0 AND **LSR**
IS UNCHANGED.

**IHOW=1** THE PROGRAM FINDS THE FIRST RECORD SUCH THAT **IR(1)** IS
GREATER THAN **IDR** AND EXITS WITH VALUE 1 AND **LSR** SET TO THE APPROPRIATE
VALUE. IF THE RECORD IS NOT FOUND **IFIND** RETURNS WITH THE VALUE 0 AND
**LSR** UNCHANGED.

**IHOW=-1** THE PROGRAM LOOKS FOR THE FIRST RECORD SUCH THAT **IR(1)**
IS LESS THAN **IDR**.

3. IGETR (IR,N,INDEX,IVALUE,IF,LOGIC)

THIS IS A SELECTIVE SEARCH FUNCTION SUB-PROGRAM THAT READS A RECORD INTO
**IR** OR NOT, ACCORDING TO VALUES OF N WORDS WITHIN THE RECORD IN THE FILE.
THE WORD NUMBERS WITHIN THE RECORD ARE SPECIFIED BY A LIST **INDEX(I), I=1,N**.
THE VALUES WITH WHICH THESE WORDS ARE TO BE COMPARED ARE GIVEN BY THE LIST
**IVALUE(I), I=1,N**. NOTE THAT ONLY FIXED POINT OR ALPHA VALUES ARE ALLOWED.
THE CONDITION OF WHETHER OR NOT A SPECIFIED WORD IN THE RECORD SHOULD BE
GREATER, EQUAL OR LESS THAN ITS CORRESPONDING VALUE IS GIVEN BY THE LIST
**IF(I),I=1,N**.

**IF (I)** SHOULD BE THE VALUE OF 1,0 OR -1 DEPENDING ON THE CONDITIONS
SPECIFIED ABOVE.

THE VALUE OF **LOGIC** SHOULD BE 1 IF ALL (LOGICAL AND) OF THE N CONDITIONS
ARE TO BE SATISFIED AND 0 IF ANY (LOGICAL OR) OF THE N CONDITIONS ARE TO BE
SATISFIED.

THE FUNCTION RETURNS WITH 1 IF THE SCANNED RECORD SATISFIES THE ABOVE
CONDITIONS AND **IR** CONTAINS THAT RECORD. IT RETURNS WITH THE VALUE 0 IF THE
RECORD DOES NOT SATISFY THE CONTITIONS.

TYPE IV WORD MANIPULATIONS

1. IUNPAK(NL,NR,IN)
THIS FUNCTION DISCARDS **NL** DIGITS ON THE LEFT OF THE WORD **IW** AND
**NR** DIGITS ON THE RIGHT OF THE WORD **IW**. THE FUNCTION RETURNS WITH THE
REMAINDER OF THE WORD RIGHT-JUSTIFIED.

2. LUNPAK(NL,NR,IW)
THIS FUNCTION IS SIMILAR TO IUNPAK EXECPT THAT IT OPERATES ON
5 ASCII CHARACTERS PER WORD AND RETURNS THE RESULTS LEFT JUSTIFIED AND
BLANK FILL TO THE RIGHT.

3. JUNPAK(NL,NR,IW)
THIS FUNCTION OPERAPTES ON ASCII WORD AND RETURNS THE
RESULTS RIGHT JUSTIFIED AS OCTAL NUMBER.

COMMENTS AT THE BEGINNING OF EACH SOURCE PROGRAM IN THE
SYSTEM DEFINE ITS USE.

THE UTILITY PROGRAMS IN THE SYSTEM ILLUSTRATE AND GIVE
PROGRAMMING EXAMPLES OF THE USE OF THE FILE MAINTAINCE SYSTEM.