File No. 360-25
GY28-2020-2
Program Logic
Version 8.1
IBM System/360 Time Sharing System
FORTRAN IV Library Subprograms
This publication describes the internal logic of the
IBM System/360 Time Sharing System FORTRAN IV
mathematical and I/O libraries, including the
mathematical, service, and I/O routines.
This material is intended for persons involved in
program maintenance, and system" programmers who are
altering the program design. It can be used to locate
specific areas of the program, and it enables the
reader to relate these areas to the corresponding
program listings. Program logic information is not
necessary for the use and operation of the program.
PREFACE
This publication is organized into six
sections and two appendixes.
Section 1 is an introduction to the
FORTRAN mathematical and I/O libraries,
including service subprograms, and provides
a brief description of their contents and
functions .
Appendix A describes those aspects of
FORTRAN data management that are unique to
the FORTRAN environment. Appendix B is a
guide to external names of FORTRAN library
routines .
PREREQUISITE PUBLICATIONS
Section 2 contains descriptions, in
figure form, of how the library routines
interact in fulfilling user requests.
Familiarity with the material contained
in the following publications is essential
to the use of this manual:
Section 3 describes each mathematical
subprogram -~ its entry names, function,
attributes , entry , exit, storage
requirement^ error checks, and (where
applicable) accuracy figures.
IBM System/360 Time Sharing System :
Concepts and Facilities , GC28-2003
IBM FORTRAN IV , GC28-2007
FORTRAN Programmer's Guide , GC28-2025
Section 4 describes the service routines
residing £n~*the mathematical and I/O
libraries — their subprograms, attributes,
entry names and entry parameters, storage
requirements, error checks, and their
operation.
REFERENCE PUBLICATIONS
Knowledge of the following publications
will be helpful in understanding the
concepts and logic of the FORTRAN Library
routines :
Section 5 describes each I/O routine -
its purpose, entry point and entry
parameters, external references, and the
details of its operation.
Section 6 gives the flowcharts of the
service and I/O routines.
IBM System/360 Time Sharing System :
FORTRAN IV Library Subprograms ,
GC28-2026
System Programmer's Guide , GC28-2008
Assembler User Macro Instructions ,
GC28-2004
FORTRAN IV Compiler PLM , GY28-2019
System Control Blocks PLM , GY28-2011
System Logic Summary , GY28-2009
Third Edition (September 1971)
Significant changes or additions to this publication will
be provided in new editions or Technical Newsletters.
This edition is current with Version 8, Modification 1 of
IBM System/360 Time Sharing System (TSS/360) and remains in
effect for all subsequent versions or modifications of IBM
System/360 Time Sharing System unless otherwise indicated.
Before using this publication, refer to the latest edition of
IBM System/360 Time Sharing System: Addendum , GC 2 8 - 2 H 3 ,
which may contain information pertinent to the topics covered
in this edition. The Addendum also lists the editions oi all
TSS/360 publications that are applicable and current.
This publication was prepared for production using an IBM
computer to update the text and to control the page and line
format. Page impressions for photo-offset printing were
obtained from an IBM 14Q3 Printer using a special print
chain*
Requests for copies of IBM publications should be made to
your IBM representative or to the IBM branch office serving
your locality.
A form is provided at the back of this publication for
reader's comments. If the form has been removed, comments
may be addressed to IBM Corporation,, System/360 Time Sharing
System Programming Publications, Department 64 3 , Neighborhood
Road, Kingston, New York 12401
Copyright International Business Machines Corporation 1969, 1970, 1971
CONTENTS
SECTION 1: INTRODUCTION • 1
SECTION 2: OVERVIEW . . . . . 2
Mathematical Library .......... 2
Subdivisions 2
Internal Relationships ......... . . 2
I/O Library ....... ...... 2
Subdivisions . . . . . * . . 3
I/O Language Control Routines 3
Data Conversion Routines ..... 3
Internal Relationships ...•••....•. .... 3
SECTION 3: MATHEMATICAL SUBPROGRAM DESCRIPTIONS 14
General Information 14
Routine Names 14
Attributes .14
Entry Parameters ...........14
Error Action 14
Exit Parameters ••..•••.••... ..14
Subprogram Summaries ........••• 14
Tables . 14
CHCBZ — Error Processor •........••.... 25
SECTION 4: SERVICE SUBPROGRAM DESCRIPTIONS 26
General Information 26
Subprogram Summaries . 26
CHCBE — interrupt and Machine Indicator Routine ........ 26
CHCIV — Dump Routine 28
CHCIW — Exit Routine 29
SECTION 5: I/O ROUTINE DESCRIPTIONS . 30
General Information •.... 30
Attributes ........... 30
Work Areas and Register Save Areas . 30
Routine Summaries . 30
CHCIA — I/O Initialization . 30
CHCIB — DCB Maintenance .31
CHCIC — I/O Control ............... 32
CHCID — NAMELIST Processor . 34
CHCIE — List Item Processor 36
CHCIF — Format Processor ..;. 36
CHCIH — Integer Output Conversion ... 37
CHCII — Real and Integer Input Conversion . 37
CHCIJ « Real Output Conversion ........ .... 39
CBCIM — Complex Input Conversion . ..... 39
CHCIN — Complex Output Conversion ...39
CBCIO — Alphameric and Hexadecimal Input Conversion ...... 39
CHCIP — Alphameric and Hexadecimal Output Conversion ...... 40
CHCIQ — Logical Input Conversion 40
CHCIR — Logical Output Conversion 40
CHCIS — General Input Conversion ......... 40
CHCIT — - General Output Conversion 41
CBCIU — ■ List Termination 41
CHCIW -- Exit 41
CHCBD — * I/O Interruption and Machine Indicator Routine 41
SECTION 6: FLOWCHARTS ........... 42
APPENDIX A: FORTRAN DATA MANAGEMENT ........ ... 89
DCB Use 89
DCB Content .89
DCB Initialization 89
Combinations of DSORG and RECFM ..•..•••89
Unformatted FORTRAN Logical Records • . • • . 90
APPENDIX B5 EXTERNAL NAMES SUMMARY ................. 91
INDEX . . . . . . • • . . • 92
ILLUSTRATIONS
Figure 1. Functional flow of mathematical library subprograms . . • 4
Figure 2. Functional flow of I/O library routines . • • 7
Figure 3. Formatted READ with list ....... 8
Figure 4. Formatted READ without list ....... 8
Figure 5. READ with NAMELIST ........ ..... 9
Figure 6. Unformatted READ with list 9
Figure 7. Unformatted READ without list . 10
Figure 8. Formatted WRITE with list . 10
Figure 9. Formatted WRITE without list .............. 11
Figure 10. WRITE with NAMELIST ...... ..... 11
Figure 11. Unformatted WRITE with list .............. 12
Figure 12. Unformatted WRITE without list . . . • 12
Figure 13. BACKSPACE, REWIND, and END FILE 13
Figure 14. EXIT and STOP ..................... 13
Table 1. I/O library calling relationships . .- 6
Table 2. Summary of directly referenced mathematical subprograms . 15
Table 3. Summary of indirectly referenced mathematical subprograms 21
Table 4. Exponentiation with integer base and exponent 24
Table 5. Exponentiation with real or double-precision base and
integer exponent 24
Table 6, Exponentiation with real or double- precision base and -
exponent .................... 24
Table 7. Entry names of indirect-reference mathematical routines . 24
Table 8. Mathematical library macro instruction . summary 25
Table 9. Summary of service subprogram characteristics, ...... 27
Table 10. Format and content of DCB prefix ............. 32
Table 11. Translation of FORMAT codes . 38
Table 12. Combinations of DSORG and RECFM values .......... 89
Table 13. External names of FORTRAN IV library subprograms 91
Chart BD. I/O interruption and machine indicator routine CCHCBD) . . 43
Chart BE. Interrupt and machine indicator routine (CHCBE) 44
Chart BZ. Error processor, math library (CHCBZ) ..... 47
Chart IA. I/O initialization CCHCIA) 48
Chart IB. DCB maintenance (CHCIB) 49
Chart IC. I/O control CCHCIC) 50
Chart ID. NAMELIST processor (CHCID) ....... 68
Chart IE. List item processor (CHCIE) • 7 ^
Chart IF. FORMAT processor CCHCIF) 71
Chart IH. Integer output conversion CCHCIH) - 74
Chart II. Real and integer input conversion (CHCII) ........ 75
Chart U. Real output conversion (CHCIJ) 76
Chart IM. Complex input conversion (CHCIM) 7 8
Chart IN. Complex output conversion CCHCIN) ..... .79
Chart IO. Alphameric and hexadecimal input conversion (CHCIO) ... 80
Chart IP. Alphameric and hexadecimal output conversion (CHCIP) ... 81
Chart IQ» Logical input conversion (CHCIQ) ............. 82
Chart IR. Logical output conversion (CHCIR) .... ..83
Chart IS. General input conversion CCHCIS) ... 84
Chart IT. General output conversion (CHCIT) . 85
Chart IU. List termination CCHCIU) ................. 86
Chart IV. Dump routine (CBCIV) .......... - 87
Chart IW. Exit routine (CHCIW) ....... 88
SECTION li
INTRODUCTION
The two FORTRAN IV libraries, mathemat-
ical and input/output CI/O) , are written in
assembler language and stored in the system
library (SYSLIB) . They are available to
all users.
Each routine in the mathematical library
is a collection of one or more subprograms.
In this publication, the term subprogram
means a routine or part of a routine that
has a single entry point and performs or
controls the performance of a single func-
tion. Mathematical library subprograms are
generally mathematical or computational in
nature, and generally return one answer
(function value) to the calling program.
Mathematical subprograms can be categorized
by use:
1. Direct reference , as in reference to
the sine subprogram in the statement
X = SIN(Y)
2. Indirect reference , as in reference to
an exponentiation subprogram in the
statement
X = Y**I
The I/O library is a group of routines
that function as a single program complex
for processing the I/O statements READ,
WRITE, PRINT, PUNCH, BACKSPACE, REWIND, and
END EILE. Processing cf READ and WRITE
statements can include list control, NAME-
LIST control, FORMAT ccntrcl, or none of
these controls. The routines within this
complex can be categorized by function:
1. Language control routines , which ana-
lyze the user's I/O request.
2. Data conversion Routines , which con-
vert data froir internal to external
form or from external to internal
form.
Both libraries also contain service rou-
tines, each of which is a collection of
subprograms. Service subprograms are
called with CALL statements or are implic-
itly called by the occurrence of certain
situations during execution. The service
subprograms in the mathematical library
handle machine exceptions and test program-
simulated machine indicators. The service
subprograms in the I/O library dump program
data onto SYSOUTs and terminate execution
cf user programs.
Each of the two libraries also contains
an error-handling routine.
Section 1: Introduction 1
SECTION 2s OVERVIEW
This section explains the subdivisions
of the FORTRAN IV mathematical and I/O
libraries and gives in figure forir an over-
view of each library. Note that the ser-
vice subprograms are divided between the
two libraries.
MATHEMATICAL LIBRARY
SUBDIVISIONS
The FORTRAN IV mathematical library con-
sists of two types of relocatable routines:
mathematical and service. Each routine
contains one or more subprograms; there is
a separate entry point for each subprogram.
Like a FORTRAN subprogram defined with a
FUNCTION statement, a mathematical subpro-
gram always returns an answer (function
value) to the calling program. Mathemati-
cal subprograms can be categorized as
direct reference or as indirect reference.
Since the user refers to direct-
reference subprograms by name, they are
explicitly called . For example, in the
statement X = SIN(Y), the user invokes the
SIN subprogram. The direct- reference math-
ematical subprograms are the logarithmic,
trigonometricr hyperbolic, square root,
absolute value, gamma function, and error
function subprograms, and some exponentia-
tion subprograms .
Since the user does not refer to
indirect- reference subprograms by name,
they are implicitly called . For example,
to invoke an exponentiation subprogram, the
user employs the exponentiation operator
(i.e., **)• The FORTRAN compiler then
causes the user program to call the appro-
priate exponentiation subprogram. To
invoke CHCBZA, the error-handling subpro-
gram, the user invokes a mathematical sub-
program without specifying the proper argu-
ments. With the exception of CHCBZA, all
indirect-reference mathematical subprograms
are exponential (compute powers of
numbers) .
The mathematical library service subpro-
grams can also be categorized as direct-
reference or as indirect- reference. The
direct-reference subprograms SLITE, SLITET,
OVERFL, and DVCHK test program- simulated
machine indicators. The indirect-reference
subprograms CHCBD1, CHCBD2, CHCBD3 , CHCBD4,
CHCBD5, and CHCBE1 handle interruptions
resulting from machine exceptions.
Each entry to an indirect-reference
mathematical subprogram (and some entries
to service subprograms) in the mathematical
library is made with two entry names:
1. Compiler-generated entry name. When
the user employs the exponentiation
operator, the compiler examines the
terms en both sides of the operator
and determines which exponentiation
subprogram to call. It then generates
an entry name that is later translated
into the system entry name.
2. System entry name. This is the name
defined by the subprogram itself, the
name of the entry point.
Each entry to a direct-reference subpro-
gram is made with the name employed by the
user. (The system entry name is the same
as the user-employed name.)
INTERNAL RELATIONSHIPS
Figure 1 gives a functional overview of
the mathematical library. The subprogram
entry names of each routine are shown under
the name of the routine. The entry names
shown for the directly referenced subpro-
grams are the user entry names. System
entry names of these subprograms are shown
in Table 7. The entry names shown for the
indirectly referenced subprograms are the
system entry names of these subprograms are
shown in Table 8. The error-processing
routine CHCBZ is not included in Figure 1.
I/O LIBRARY
Since the user has 10 entry points to
the FORTRAN IV I/O library, this library
can be thought of as 10 subprograms and a
number of subroutines for these subpro-
grams. The I/O library can also be
| described as a group of 21 routines that
interact in various ways, depending upon
the user's request. Except in "Section 4:
Service Subprograms," this publication
describes the I/O library in terms of its
routines •
Three of the ten entry points to the I/O
library are for processing the I/O state-
ments READ, WRITE, REWIND, BACKSPACE, END
FILE, PRINT, and PUNCH. Processing of READ
or WRITE statements can include list con-
trol, NAMELIST control, FORMAT control, or
none of these controls. The other seven
entry points are to service subprograms, in
two routines, that execute the statements
STOP, PAUSE, CALL DUMP, CALL PDUMP, and
CALL EXIT, and act upon requests by other
library routines for termination of program
execution.
SUBDIVISIONS
There are two groups of I/O routines :
I/O language control routines and data con-
version routines. These groups interact,
in fulfilling an I/O request, by means of a
common communication and work region.
I/O Language Control Routines
There are three types of I/O language
control routines: I/O operation control,
I/O list control, and I/O services control.
These routines analyze the user's I/O
requests to determine information such as:
the type of I/O operation to be performed;
the number and type of list items present,
if any? the type of NAMELIST or FORMAT con-
trol, if any; and the I/O statement rela-
tionships with a user-specified DDEF
command.
I/O OPERATION CONTROL ROUTINES : These rou-
tines control the I/O request by creating,
if necessary, a data control block (DCB) f
and analyzing FORMAT and NAMELIST control
specified by the user. After this informa-
tion is processed, the I/O operation con-
trol routines interface with the TSS/360
data management routines that actually ful-
fill the I/O request. The interface with
data management is accomplished by the rou-
tines CHCIB and CHCIC, via the data manage-
ment macro instruction facilities.
I/O LIST CONTROL ROUTINES : These routines
examine the list items, if any, in each I/O
request to determine the type of conversion
to be performed. After the type of data
conversion is determined, control is given
to the I/O operation control routines which
in turn call the appropriate data conver-
sion routines for final processing.
I/O SERVICES CONTROL ROUTINES : These rou-
tines are responsible for creating user-
requested dumps, initiating a pause in the
user program, and terminating a user pro-
gram as a result of either normal proces-
sing completion or program errors. I/O
services control routines also interact
with the I/O operation control and with the
group of data conversion routines for the
preparation of the user's data in virtual
storage for subsequent output on SYSOUT.
Data Conversion Routines
The data conversion routines are subdi-
vided into routines used for input proces-
sing and routines used for the preparation
of output. These routines can process all
the permissible types of FORTRAN-formatted
data specified in either a FORMAT or NAME-
LIST statement.
When converting a user's data, the data
conversion routines interact with each
other according to the requirements of the
user-specified FORMAT or NAMELIST control.
For example, for input data that is defined
by a G- format conversion code. General
Input Conversion (CHCIS) is called. This
routine analyzes the data to determine
whether it is integer, real, logical, or
alphameric and calls the appropriate data
conversion routine.
INTERNAL RELATIONSHIPS
Figure 2 gives a functional overview of
the I/O library.
Table 1 shows the calling relationships
between the user program, the FORTRAN I/O
library routines, Data Management, and the
Supervisor.
Since the routine functions and interre-
lationships vary, depending on the kind of
I/O operation being performed, separate
diagrams (Figures 3-14) describe the dif-
ferent kinds of I/O operations. Exceptions
to the logical flows shown in Figures 3-14
are covered under the individual routine
descriptions in "Section 5: I/O Routine
Des cript ions . *
The types of I/O operations and their
figure references are:
Type of Operation (Function)
Formatted READ with List
Formatted READ without List
READ with NAMELIST
Unformatted READ with List
Unformatted READ without List
Formatted WRITE with List
Formatted WRITE without List
WRITE with NAMELIST
Unformatted WRITE with List
Unformatted WRITE without List
REWIND, BACKSPACE, and END FILE
EXIT and STOP
Figure
Reference
Figure
3
Figure
4
Figure
5
Figure
6
Figure
7
Figure
8
Figure
9
Figure
10
Figure
11
Figure
12
Figure
13
Figure
14
Section 2: Overview 3
ARCTANGENT
r
I
r~"
MATHEMATICAL
SERVICES CONTROL
INTERRUPT AND
CHCBE*
MACHINE INDICATOR
ROUTINE
CHCBE2
CHCBE3
CHCBE4
CHCBE5
OVERFL
DVCHK
SLITE
SLITET
I
CHCBR
DATAN
DATAN2
CHCBQ
ATAN
ATAN2
1
I J
COMMON & NATURAL LOGARITHM
TRIGONOMETRIC
TANGENT
L.
C HCAZ
DTAN
DCOTAN
CHCAY
TAN
COTAN
Figure 1. Functional flow of mathematical library subprograms (page 1 of 2)
GAMMA;
LOG-GAMMA
U.OC
J
REAL OR INTEGER
BASE TO REAL
POWER
WBT
GAMMA
ALGAMMA
L_.
CHCSV
DGAMMA
DLGAMMA
1
ERROR FUNCTION;
COMPLEMENTED
ERROR FUNCTION
J
r
CHCBU
ERF
ERFC
I
CHCBW
DERF
DERFC
. I
HYPERBOLIC TANGENT
CHCAK
TANH
CHCAL
DTANH
L..
J
EXPONENTIAL
INTEGER BASE
CHCBG*
TO INTEGER
CHCBGA
POWER
CHCBG B
CHCBGC
CHCBGD
REAL BASE TO
INTEGER POWER
COMPLEX BASE
TO INTEGER
POWER
CHCBH*
CHCBHA
CHCBHB
CHCBC*
CHCBCA
CHCBCB
TRIGONOMETRIC
SINE & COSINE
r~ ' 1
CHCAQ
CSIN h
CCOS
CHCAI
SIN
COS
L_T
CHCAR
CDSIN
CDCOS
CHCAJ
DSIN
DCOS
SIN) (COS) (DSIN) foCOS
J
HYPERBOLIC
SINE & COSINE
r
CHCBA
SINH
COSH
CHCBB
DSINH
DCOSH
"i
5 LOG)
4
CHCBK*
CHCBKA
CHCBKB
CHCBKC
CHCBKD
CHCBKE
CHCBI*
CHCBIA
CHCBIB
CHCBM*
CHCBMA
CHCBMB
REAL DOUBLE
PRECISION BASE
TO REAL DOUBLE
PRECISION POWER
DOUBLE PRECISION
BASE TO INTEGER
POWER
COMPLEX, DOUBLE
PRECISION BASE TO
INTEGER POWER
Note: Asterisk (*) denotes
that following subprograms
are implicitly called
(Indirect reference).
i_.
_f
Figure 1. Functional Flow of Mathematical Library Subprograms (Page 2 of 2)
Section 2: Overview 5
Table 1. I/O library calling relationships
n. Routines
\. Called
Calling \
Routines \^
<
u
X
u
CO
u
X
u
u
X
u
9
u
X
u
u
u
U
X
u
u
X
u
u
X
u
"—5
u
X
u
1
u
X
u
z
u
X
u
g
u
X
u
u
X
u
a
u
X
u
as:
U
X
u
U
X
u
u
X
u
u
X
u
>
u
X
u
u
X
u
o
QD
u
X
u
c
E
a*
2 ?
Q 2
o
Ql
3
USER
PROGRAM
X
X
X
X
X
X
CHCfA
X
X
X
X
X
CHCIB
1 [ "
X
X
X
CHC1C
X
X
X
X
CHCID
X
X
CHCIE
X
X
X
CHCIF
X
X
X
X
X
X
X
X
X
X
X
X
X
CHCIH
CHCII
— - T i "i "
CHC1J
'
CHCIM
|
X
CHC1N
X
CHCIO
—
CHCIP
CHCIQ
CHCIR
"■ f ~
I
CHCIS
X
X
X
CHCIT
X
X
X
CHCIU
X
X
CHCIV
>
X
x
X
X
X
X
X
CHCIW
X
X
[
X
CHCBD
i
G'j
O
o
<
ft
o
0)
o
70
o
I/O tongue Con fro f Routines
U*er Program
DUMP, POUMP, READ, WRITE, PRINT,
EXIT, STOP, . PUNCH, BACKSPACE, List I
REWIND, END FILE I
PAUSE
Data Conversion
Routines
Input
r
General
(CHCIS)
Integer and
Real
(CHCII)
Complex
(CHCIM)
Alphameric and
Hexadecimal
(CHCIO)
Output
General
(CHCIT)
integer
(CHCIH)
Real
(CHCIJ)
Complex
(CHCIN)
Logical
(CHCIR)
Alphameric and
Hexadecimal
(CHCIP)
User
Program
m _
7
i
t
T ____, '
_i
1
1
t V
A
♦ I
CHCIA
I/O Initial-
ization
CHCSE
List Processing
Enter for each
item
1
1
I
2
1
t
J
\
Is f
1
i
1
1
1 3
1
\ 1
if 1
i
<
1
p I
CHCIF
FORMAT Processor
process each item
according to
FORMAT speci-
fication
1 •
CHCIB
Find a DCB
defining associ-
ated data set
1
r
i
—
CHCIU
List
Termination
CHCIC
Read o logical
record
1
i
I
i
_i
5A
-.~r~
CHCII, IM, IO,
jQ, IS
Appropriate data
conversion routine
for each item
Read sub-
sequent records
if necessary
Figure 3. Formatted READ With List
User
Program
i
t
1
f
r~ -
i
—
1
1
CHCIA
I/O Ini
ization
hial-
4
CHCIF
FORMAT Processor
move data items
from buffer into
FORMAT
m .
t
♦
2
3
„ ... .1
r~ -
f i
—
1
_l
1
L
1
1
I
CHCIB
Find a DCB
defining associ-
ated data set
CHCIC
Read a foglca I
record
Read
subsequent
record if
necessary
Figure 4. Formatted READ Without List
User Program
■ !
2
CHCtA
1 /O Initialization
4
_ 1
1
1 1
"I
1
CHCIB
Find a DCB
3
1
1
1 1
CHCID
NAMELIST
Processor
i
J5A
1
1
_J
CHCIC
Read a logical
record
•..!
CHCU, IM,
iO, IQ # IS
Appropriate
data conversion
routine
Read
subsequent
records if
necessary
Figure 5, READ With NAMELIST
User Program
"~l
"1
| 5
|
I
1
t
i i
4
1
1
1 1
CHCIA
I/O Initialization
CHCIU
List Termination
1
2
i
1
1
1
' 1
3
t
1
- 1
, 1
CHCI8
Find a DCB
CHCIC
Read a logical
record
CHCiE
List Processor
{Enter for
each item)
r
1 4A
J
Read
subsequent
records If
necessary
Figure 6. Unformatted READ With List
Section 2: Overview 9
User Program
CHCIA
I/O Initialization
CHCIB
Find a DCB
Figure 7. Unformatted READ Without List
CHCIC
Read a logical
record
L
CHCIB
Find a DCB
defining asso-
ciated data set
User
Prograrr
1
CHCiA
I/O Initial
ization
L
CHCIC
Initialize pointers
to buffer in which
to construct data
record
Write buffer in
output data set,
when ful I
1
±
CHCJE
List Processingsenter
for each item f pyf-
list in buffer
r
T
JZ.
C HCIF
FORMAT Processor
process each item
according to
FORMAT speci-
fication
6A
-A L_
CHCIU
List
Termination
_1_
CHCIhl, U_, IN, U>,
NR^ IT
Appropriate data con-
version routine for
each item in buffer
_l
Figure
Formatted WRITE with
..ist
10
1
User
Program
t
f
1
_ -
1
1
1
I
4
CHCIF
FORMAT Processor
move data items
into buffer from
FORMAT
CHCIA
I/O Initio
ization
f
t
I
"3
♦
1
f 1
1
-J
i
1
I
1
»
1
5
CHCIB
Find a DCB
defining associ-
ated data set
CHCIC
Initialize pointers
to buffer in which
to construct data
record
Write buffer
output data
when full
In
set,
Figure 9. Formatted WRITE Without List
User
Program
1
1
t
1
» I
2
CHCIA
I/O Initial-
ization
4
'
'
i i
i
i
i
CHCIB
Find a DCB
3
1
t
i
CHCID
NAMELIST
processor
r~~
CHCIC
Initialize pointers
to buffer for data
record construction
i
5
\
i
ISA
CHCIH, |J, IN,
ify IBxil
Appropriate data
conversion routine
for each item
Write buffer in
output data set,
when full
—
Figure 10 • WRITE With NAMELIST
Section 2: Overview 11
r
User Program
'~l
4
1
1
I
i
1 ♦
L
J
1 6
1
W 1
T 1
CHCIA
f/O Initial-
ization
CHCIE
List Processor
(enter for each
item)
CHCIU
List Termination
insure last
record is written
2
♦ 1
1
I
t
1
I
1
l
_ J
t
'
r " "
_J
CHCJC - Initialize
pointers to buffer
in which to con-
struct data record
5 1
1
1
1
1
7
" """ "'
CHCIB
Find a DCB
Write buffer in
output data set,
when full
1
m
Figure 11. Unformatted WRITE With List
User Program
s
_! 4
1
! I
1
CHCIA
I/O Initial-
ization
- 1
1
1
1
U
1
1
1
t
1
L
3
!
2
I
* 1
1
I
r 1
CHCIC - initial-
ize pointers to
buffer in which
to construe t
dummy record
CHCfB
Find a DCB
Write buffer
consisting of
two bytes of
zeros
.
_l
Figure 12- Unformatted WRITE Without List
12
\
User Program
V 3
t
i i
1
CHCIA
I/O Initial-
ization
CHCIC - Perform
REWIND and
BACKSPACE opera-
tion. END FILE is
performed on subse-
quent WRITE
CHCIB
Find a DCB
Fiorure 13. BACKSPACE f RFWIND, and END FILE
2
3
User Program
1
1
r
CHCIB
Find all DCBs
currently open
CHCIC
For each open
DCB found by
CHCIB, the
DCB is closed
CHCIW
Exit Routine
a» — .-_
4
i
f Exit to Com- \
I mand System J
Figure m. EXIT and STOP
Section 2: Overview 13
SECTION 3: MATHEMATICAL SUBPROGRAM DESCRIPTIONS
This section gives the following infor-
mation on each mathematical library subpro-
gram # excluding the service subprograms,
which are described in Section 4:
Directly referenced subprograms. The
order is the same as that in the list
of operands within the parentheses in
the corresponding FORTRAN source state-
ment. For example the source statement
Maine
Name of containing routine
Function
Attributes
Entry
Exit parameter (function value)
Storage requirement
Error check
Accuracy figures (where applicable)
Since mathematical subprograms perform
standardized computations, a detailed
description of operation is given only for
CHCBZ, the error-handling routine. The
algorithms of direct-reference mathematical
subprograms are described in FORTRAN IV
Library Subprograms , GC28-2026.
GENERAL INFORMATION
Certain information is common to all
mathematical routines and their subpro-
grams. This information includes;
• Routine names
• Attributes
• Entry parameters
® Error action
• Exit parameters
Routine Names
All mathematical library routines have
five-letter names beginning with ' CHCA* or
f CHCB f .
Attributes
All mathematical library routines use
type-I linkage and are nonprivileged*
reenterable r and closed.
ANS=SIN( RADIAN)
in FORTRAN coding corresponds to an
assembler-language call containing one
address in the parameter list — the
address of RADIAN. The FORTRAN
statement
ANS=ATAN2(X, Y)
produces a linkage with a parameter
list containing the addresses of X and
Y, in that order. The assembler lan-
guage programmer's linkage to ATAN2
must do the same.
• Indirectly referenced subprograms. The
order for the exponentiation subpro-
grams is s address of the number to be
raised to a power and the address of
the power itself.
Error Action
All mathematical subprograms that check
for error call CHCBZ upon finding such
error. CHCBZ then prints an error message
and terminates execution. See the descrip-
tion of CHCBZ at the end of this section.
Exit Parameters
All subprogram results are returned in
registers, as follows?
Integer - General register
Real - Floating register
Complex - Real part in floating register
f complex part in floating
register 2
SUBPROGRAM SUMMARIES
Entry. Parameters
Each mathematical subprogram receives
one or two arguments from the calling pro-
gram, in the form of a parameter list
pointed to by register 1. The parameter
list must contain the addresses of the
arguments in the proper order:
TABLES
Tables 2 and 3 give the following infor-
mation concerning the mathematical
subprograms:
FUNCTION ; A brief description of the type
of mathematical operation performed.
14
I Table 2. Summary of directly referenced mathematical subprograms (page 1 of 3)
I
2
3
4
5
6
7
8
9
10
Function
Entry
Name
Definition
Arguments)
Function
Value
Returned
Error Condition
Storage
Estimates
Other Subprograms Required
Routine
Name
Accuracy Figures
Argument
Range
Sample
E/U
relative
absolute
No,
Type
Range
Hex
Dec
M(«)
<r(t)
m(e)
*(E)
COMMON AND
NATURAL
LOGARITHM
CDLOG
Ln (arg) or Log e (org)
See Note 8
1
COMPLEX * 16
arg # + Oi
COMPLEX * 16
Argument -0 + 0!
1E8
488
CDABS, DLOG, DATAN2, DSQRT
CHCAP
The full range
except (1 + Oi)
Note 1
2-72 x 10' 16
5.38x 10"' 7
CLOG
Ln (arg) or Log e (arg)
See Note 8
1
COMPLEX * 8
arg * + Oi
COMPLEX * 8
Argument » + Oi
1D0
464
CABS, ALOG, ATAN2, SORT
CHCAO
The full range
except (1 + 0?)
Note 1
7.15x 10" 7
1.36x 10" 7
DLOG
Ln (arg) or Log e (arg)
1
REAL * 8
arg >
REAL * 8
Argument S
21A
538
CHCAF
0.5 £ X £ 1.5
U
4.60 x 10" 17
2.09 x 10" 17
X<0.5, X > 1.5
E
3.32x 10" 16
5.52x 10" 17
DLOG10
Log 1Q (arg)
1
REAL* 8
arg >
REAL * 8
Argument 5
21A
538
CHCAF
0.5 i X£ 1.5
U
2.73x 10" 17
l.07x 10"
X<0,5, X > 1.5
E
3.02 x 10' 16
6.65 x 10~ 17
ALOG
Ln (arg) or Log e (arg)
1
REAL * 4
arg >
REAL * 4
Argument S
1D0
464
CHCAE
0.5 i X <1.5
U
6.85 x 10" 8
2.33 x 10' 8
X< 0.5, X >1.5
E
8.32 x 10" 7
1.19x 10" 7
ALOGIO
Log 10 (org)
1
REAL * 4
arg >
1D0
464
CHCAE
0.5 i X $1.5
U
7. 13 x 10" 8
2.26 x 10* 8
REAL * 4
Argument 5
X< 0.5, X > 1.5
E
1.05 x 10" 6
2.17x 10" 7
EXPONENTIAL
CDEXP
e arg
1
COMPLEX * 16
reol arg < 174.673
|imag arg| < 2 w
COMPLEX * 16
Real Argument > 174.673
i 50
pmaginary Argumentj 2t 2
2710
624
DEXP, DSIN, DCOS
CHCAN
|X,| < 1, |X2)S |
U
3.76x !0" 16
l.lOx io" 16
|X,|<20, |X2J ^ 20
U
2.74 x 10~ 15
9,64x 10" 16
CEXP
e arg
1
COMPLEX * 8
real arg 174,673
Jimag arg| < 2 7f
COMPLEX * 8
Real Argument > 174.673
(imaginary Argument) i 2
250
592
EXP, SIN, COS
CHCAM
N<,70,|X2|Sf
U
9.93 x 10" 7
2.67 x 10~ 7
|X,| S 170,
! <|*$|£20
u
1.07x 10" 6
2.73 x 10" 7
DEXP
e ar 9
1
REAL * 8
arg < 174,673
REAL * 8
Argument > 174,673
2C0
704
CHCAD
|x|Si
u
2.04x 10* 16
5.43x 10" 17
1 < |x| £20
U
2.03 x 10' 16
4.87x 10~ 17
20 < | X|$170
u
1.97x 10* 16
4.98x 10" 17
EXP
e °*9
1
REAL * 4
arg < 174.673
REAL • 4
Argument > 174,673
1A8
424
CHCAC
|x|ii
U
4.65 x I0" 7
128x 10" 7
1< |X| i 1 70
U
4.42 x 10" 7
1.15 x 10" 7
SQUARE ROOT
CDSQRT
(arg) or ^J arg
1
COMPLEX * 16
Any
COMPLEX • 16
None
148
328
CDABS, DSQRT
CHCAT
The full range
Note 1
1.76* 10" 16
4.06x 10" 17
CSQRT
(arg) or *J arg
1
COMPLEX * 8
Any
COMPLEX * 8
None
138
312
CABS, SORT
CHCAS
The full range
Note 1
7.00x 10" 7
1.71 x 10" 7
DSQRT
(arg) or *J arg
I
REAL * 8
arg #0
REAL * 8
Negative Argument
160
352
CHCAB
The full range
E
1.06x 10" 16
2-16x 10"' 7
SORT
(arg) ' or -/ arg
1
REAL * 4
arg #0
REAL * 4
Negative Argument
158
344
CHCAA
The full range
E
4.45 x 10" 7
8.43 x 10" 8
ARCSINE
AND
ARCCOSINE
DARSIN
arcsine (arg)
1
REAL * 8
arg < 1
REAL * 8
Argument > 1
288
648
DSQRT
CHCAX
-l £ x <+i
U
2.04 xlO" 16
5.15x 10" 17
DARCOS
arecosine (arg)
1
REAL * 8
arg £ 1
REAL * 8
Argument > 1
288
648
DSQRT
CHCAX
-i < x S + i
U
2.07x 10" 16
7.05 x 10' 17
ARSIN
arcsine (arg)
1
REAL * 4
largl <1
REAL * 4
JArgumentj > 1
1F0
496
SORT
CHCAW
-i < x S+i
U
9.34 x 10" 7
2.06x 10* 7
ARCOS
arecosine (arg)
1
REAL * 4
|arg| <1
REAL * 4
Argumenr > 1
1F0
496
SQRT
CHCAW
-1 < X < + 1
u
8.85 x 10" 7
3.19x 10" 7
ARCTANGENT
DATAN
arctan (arg)
1
REAL * 8
Any
REAL * 8
None
288
648
CHCBR
The full range
Note 7
2.18x 10" 16
7.04x 10" 17
DATAN2
arctan (argi/arg2)
2
REAL * 8
arg *0
REAL * 8
x, = x 2 =
288
648
CHCBR
The full range
Note 7
2.18 x 10" 16
7.04 x IO* 17
ATAN
arctan (arg)
1
REAL * 4
Any
REAL * 4
None
1E8
488
CHCBQ
The full range
Note 7
1.01 x 10* 6
4.68 x 10* 7
ATAN2
Cretan (arg,/arg„)
2
1
REAL * 4
arg *0
REAL * 4
Xl -x 2 =o
1E8
488
CHCBQ
The full range
Note 7
1.01 x 10" 6
4.68 x 10" 7
CDSIN
sin (arg), arg in
radians
COMPLEX * 16
|realorg| < T^ »
limag argl < 174,673
COMPLEX * 16
|Real Argument] > 2 5 * w
Imaginary Argument > 174,673
340
832
DSIN, DCOS, DEXP
CHCAR
|Xl|*io,|x 2 |Sl
U
2.35 x 10" 15
See Note 4
2.25 xlO" 16
TRIGONOMETRIC
SINE & COSINE
CDCOS
cos (arg), arg in
radians
1
COMPLEX * 16
|realarg| < 2 50lr
|imag arg| < 174,673
COMPLEX * 16
|Real Argument] * 2 ^ w
Imaginary Argument > 174,673
340
832
DSIN, DCOS, DEXP
CHCAR
N - ]0 >\*2\- ]
U
3.98 x 10" 15
See Note 3
2.50 x 10" 16
CSIN
sin (arg) , arg in
radians
1
COMPLEX * 8
|real arg| < 2 18 »
imag org < 1 74,673
COMPLEX • 8
|Rea! Argument] > 2 18 » 2F8
Imaginary Argument > 174,673
. . i
760
SIN, COS, EXP
CHCAQ
|X,| S I0,|X2|S1
U
1.92 x 10" 6
See Note 6
7.38 x 1Q~ 7
Section 3; Mathematical Subprogram Descriptions 15
I Table 2. Summary of directly referenced mathematical subprograms (page 2 of 3)
1
2
3
4
5
6
7
8
9
10
Function
Entry
Nome
Definition
Argumenf(s)
Function
Value
Returned
Error Condition
Storage
Estimates
Other Subprograms Required
Routine
Accuracy Figures
A
Sample
E/U
Relative
Absolute
No.
Type
Range
Hex
Dec
Range
M (f)
<r( «)
M (E)
»(E)
TRIGONOMETRIC
SINE & COSINE
(Continued)
CCOS
cos (org), org in
radians
1
COMPLEX * 8
i 1 I < o 18 »
real arg ** /
imag arg < 1 74,673
COMPLEX * 8
JReal Argument 2 2 18 »
Imaginary Argument > 174.673
2F8
760
SIN, COS, EXP
CHCAQ
|x,|Sio,|x 2 |< i
U
2.50 x IO" 6
See Note 2
7.66 x IO" 7
DSIN
sin (org), arg in
radians
1
REAL * 8
1 1 < -)50 ,
arg * 2
REAL * 8
[Argument! > 2 50 *
2B8
696
CHCAJ
l*l*f
U
3.60x IO" 16
4.82 x IO* 7
7.74 x IO" 17
1.98 x IO** 17
I<|X|<10
U
1.64x IO" 16
6.49 x IO** 17
10 < | X | < 100
U
2-68 x IO" 15
1.03 x IO" 15
DCOS
cos (org), arg in
rodians
1
REAL * 8
|arg| < 2 50 »
REAL * 8
Argument - 2 *
2B8
696
CHCAJ
< x< »
U
1.79x IO" 16
6.53 x IO" 17
-10 i X <
w < X < 10
U
1.75 x IO" 16
5.93 x IO" 17
io < | x| S ioo
U
2.64 x IO" 15
1.01 x io" 15
SIN
sin (arg), arg in
rod i am
1
REAL * 4
|arg| < 2 18 tr
REAL * 4
Argument t 2 w
1F8
504
CHCAI
1*1*1
U
1.32 x IO" 6
1.82x IO" 7
1.18x IO" 7
4.55 x 10~ 8
* <|x|Sio
10 < | x| - 100
U
1.15 x IO" 7
4.64 x 10" 8
U
1.28 x 10~ 7
4.52 x IO** 8
COS
cos (arg), arg in
radians
1
REAL * 4
|org| <2 18 "
REAL * 4
1 Argument 1 > 2 18 *
1F8
504
CHCAI
OiXif
U
1.19x IO" 7
4.60 x IO* 8
-10 S X <
» < X £ 10
U
1.28 x IO" 7
4.55 x IO" 8
10 < | x | Sioo
U
1.14 x IO" 7
4.60 x IO* 8
TRIGONOMETRIC
TANGENT
DTAN
ran (arg), arg in
radians
I
REAL * 8
|«rfl| < 2 50 *
REAL * 8
1 Argument > 2 50 *
Argument too close to a Singularity
(i.e. , too close to an odd
multiple of?r/2)
2F8
760
CHCAZ
1*1*1
;<|x|s«
U
3.41 x IO" 16
6-27 x IO** 17
U
-12
1.43x 10 l£
See Note 5
2.95 x IO" 14
f <|x|Sio
U
2.78x IO* 13
See Note 5
7.23x IO* 15
io <| x|Sioo
U
-12
3.79 x 10 {i
See Note 5
9.50x 10** 14
OCOTAN
cofan (arg), arg in
rod ions
1
REAL * 8
|arg| < 2 50 *
REAL * 8
i 1 50
Argument it 2 v
Argument too close to a Singularity
(i.e. , too close to a
multiple of w)
2F8
760
CHCAZ
1*1* J
U
2.46x IO" 16
See Note 5
8.79 x IO" 17
;<|x|S f
U
2.78 x 10" 13
See Note 5
8.61 x IO** 15
f <| x | S io
U
5.40x 10** 13
See Note 5
1
1.13x 10" M
to <| x| S 100
U
-13
8.61 x 10
See Note 5
4.61 x IO" 14
TAN
tan (arg), arg in
radians
1
REAL * 4
|arg| < 2 18 "
REAL * 4
1 Argument 1 > 2 18 *
Argument too close to a Singularity
(i.e. , too close to an odd
multiple of it/2)
288
648
CHCAY
1*1*4*
U
1.71k IO* 6
2.64 x 10* 7
J<l*l s 5
U
1.05 x IO" 6
See Note 5
3.59 x 10" 7
J <|x|Sio
U
6.49 x IO" 6
See Note 5
3.38 x IO" 7
io<| x | S ioo
u
1.57x IO" 6
See Note 5
3.07 x IO** 7
COTAN
cotan (arg), arg in
radians
1
REAL * 4
1 1 < o 18 *
arg * 2
REAL * 4
Argument 2 2 *
Argument too close to a Singularity
(i.e. , too close to a
multiple of it)
648
CHCAY
|x|S f
u
1.07x IO" 6
3.58 x IO" 7
J<l*l*5
u
1.40x IO** 6
See Note 5
2.56 x 10" 7
f <|x|Sio
u
1.30x IO" 6
See Note 5
3.11 x IO -7
10 < | X | ^ 100
u
1.49 x IO" 6
See Note 5
3.15x IO** 7
HYPERBOLIC
SINE & COSINE
DSINH
sinh (arg)
1
REAL * 8
|org| < 175.366
REAL * 8
Argument 2 174,673
250
592
DEXP
CHCBB
|x| - 0.88137
u
2.06 x IO" 16
3.74 x IO** 17
0.88137 <| x|£ 5
u
3.80 x IO** 16
9.21 x 10" 17
DCOSH
cosh (arg)
1
REAL * 8
|arg|< 175.366
REAL * 8
Argument 1 174,673
250
592
DEXP
CHCBB
-5 £ X £ +5
u
3.63 x 10* 16
9.05 x 10*"' 7
SINH
sinh (org)
1
REAL * 4
|arg| < 175.366
REAL * 4
Argument t 174,673
1F8
504
EXP
CHCBA
-5 S X < + 5
u
1.26 x IO" 6
2.17x 10** 7
COSH
cosh (arg)
1
REAL * 4
|arg|< 175.366
REAL * 4
1 Argument 1 > 174,673
1F8
504
EXP
CHCBA
-5 < X < +5
u
1.27 x IO** 6
-7
2.63 x 10
Section 3: Mathematical Subprogram Descriptions 17
I Table 2. Summary of directly referenced mathematical subprograms (page 3 of 3)
1
2
3
4
5
6
7
8
9
10
Function
Entry
Definition
Argument(s)
Function
Value
Returned
Error Condition
Storage
Estimates
Other Subprograms Required
Routine
Name
Accuracy Figures
Argument
Range
c , | relative
absolute
No.
Type
Range
Hex
Dec
Sample i
E/U M( °
*(«)
M (E)
»(E)
HYPERBOLIC
TANGENT
DTANH
tanh (org)
1
REAL * 8
Any
REAL * 8
130
304
DEXP
CHCAL
|x| S 0.54931
U | 1.91 , 10"' 6
3.86x 10" 17
0.54931 < |x| i 5
U 1 1.54k 10" 16
1.87, I0 _1/
TANH
tanh (org)
1
REAL * 4
1 Anv
REAL * 4
164
356
EXP
CHCAK
|x|< 0.7
U 1 8.48 x 10~ 7
1.48x 10"
0.7< |x| <5
U 2,44 x 10 -7
4.23x 10~ 8
ABSOLUTE
VALUE
CDAB5
hi
1
COMPLEX * 16
Any
See Note 9
REAL * 8
C8
200
DSQRT
CHCAV
The full range
Note 1
2-03x 10~ 16
4.83 x 10* 17
CABS
ar 9
1
COMPLEX * 8
Any
See Note 9
| REAL * 4
CO
192
SORT
CHCAU
The full range
Note 1
9.15 x 10 _/
2.00 x 10" 7
ERROR
FUNCTION
ERF
2 /"V^du
1
REAL * 4
Any
1 REAL * 4
208
520
EXP
CHCBU
M<i
U
8.16x 10* 7
1 . 1 x 1 o" 7
I < jxj < 2.04
U
1 13.x 10~ 7
3.70 x 10" 8
2.04 < |x|< 3. 91 oo
U
5.95 x 10" 8
3.41 x 10" 8
DERF
* iy«
1
Real * 8
Any
Real * 8
328
808
DEXP
CHCBW
•mTi
U
!.89x 10" 16
2.60x 10" 17
1 < |x| S 2,04
u
2.87* 10" 17
9.84 x 10" 18
2.04 <|x| < 6.092
u
1.39* 10" 17
8.02x 10"' 8
COMPLEMENTED
ERROR FUNCTION
ERFC
1 -erf (x) or
if .^
Vtr x
1
REAL * 4
Any
| REAL * 4
|
1 1
208
520
EXP
CHCBU
-3.8 < X <
u
9.10x 10" 7 j 2-96x 10*" 7
£ X < 1
u
7.42x 10" 7
1.27x 10" 7
1 < Xi 2.04
u
1.54 * 10" 6
-7
3 7R y 10
2.04 < X< 4
u
2-28 x 10~ 6
3.70 x 10"
4 < X < 13,3
u
1.55 x 10" 5
8.57x 10" 6
DERFC
1 - erf (x) or
— 1 -u du
-Jit J%
1
Real * 8
Any
328
808
DEXP
CHCBW
-6 < X<
u
2.08 x 10" 16
6.52x 10" 17
Real * 8
i X i 1
u
1.40x 10" 16
2.59x 10" 17
i < X !j ■/. 04
u
4.11 x Uf 16
8.86 x 10" 17
2,04 < X<4
u
r 3.26x 10" i6
H.65 , 10" 1 '
4 < X < 13.3
u
3.51 x 10" 15
1.96x 10" 15
GAMMA (r)
GAMMA
1 u e du
J
1
REAL * 4
X> 2~ 252
X < 57.5744
REAL • 4
Real Argument > 57,5744
Real Argument < 2" 252
350
848
EXP, ALOG
CHCBT
< X < 1
u
9.86 x 10~ 7
3.66 x 10" 7
1 -: X < ; 2
u
1 .13 x 10* 7
3.22x 10* 8
2 < X < 4
u
9.47 k !0" 7
3.79x 10" 7
4 < X < 8
u
2-26x 10" 6
8-32 x 10"''
8 < X < 16
u
2-20x 10* 5
7.6! x 10" 6
16 < X < 57
u
4-62 x 10" 5
1.51 x 10" 5
DGAMMA
^0
1
REAL* 8
X > 2" 252
X < 57.5744
REAL * 8
Real Argument > 57.5744
Real Argument < 2" 252
420
1 056
DEXP, DLOG
CHCBV
< X < 1
u
2.14x 10"' 6
7.84 x 10" 17
1 < X < 2
u
2.52a 10" 17
6.07 x 10"' 8
2 < X < 4
u
2.21 x 10" 16
8-49x 10" 17
4 < X < 8
u
5.05 x !0" 16
1 . 90 x 10
8 < X < 16
u
6.02 x 10" 15
1.78 x 10" 15
16 < X < 57
u
-14
1 - 16 x 10
4.11 x 10" 15
LOG - GAMMA
ALGAMA
/*°° x-1 -u ,
log e / u e du
1
REAL * 4
X >
X < 4,2913 x 10 73
REAL * 4
Real Argument > 4.2937 x IQ 73
Real Argument <
350
848
EXP, ALOG
CHCBT
< X < 0.5
u
1.16x 10' 6
3.54 x 10" 7
0.5 < X < 3
u
9.43 x 10~ 7
3.42 x 10' 7
3 < X < 8
u
1.25 x 10~ 6
3.04 x 10" 7
8 < X < 16
u
1.18 x 10" 6
3.80 x 10" 7
16 < X < 500
u
9.85 x 10" 7
1.90x 10" 7
DLGAMA
log e / u e du
I
REAL * 8
X >
X < 4.2913 x I0 73
REAL * 8
73
Real Argument > 4,2937 x 10
Real Argument <
420
1 056
DEXP, DLOG
CHCBV
< X £ 0.5
u
2.77x 10~ 16
9.75 x 10" 17
0.5 < X < 3
u
2.24 x 10* 16
7.77x 10" 17
3 £ X < 8
u
2.89x 10" 16
8.80x 10"' 7
8 S X < 16
u
2.86x 10" 16
8.92x 10" 17
16 £ X < 500
u
1.99x 10" 16
3.93 x 10" 17
Notes 1 .
2.
The distribution of sample arguments upon which
these statistics are based is exponential radially
and is uniform around the origin.
The maximum relative error cited for the CCOS
function is based upon a set of 2000 random
arguments within the range. In the immediate
proximity of the points (n + 1/2)* + Oi (where
n = 0, ±1 , ±2, . . . ,) the relative error can be
quite high, although the absolute error is small.
3
4
The maximum relative error cited for the CDCOS
function is based upon a set of 1500 random arguments
within the range. In the immediate proximity of the
points (n + ]/2)ir + 0i (where n = 0, ± 1 , ±2, . . . ,)
the relative error can be quite high although the
absolute error is small.
The maximum relative error cited for the CDSIN
function is based upon a set of 1500 random arguments
within the range. In the immediate proximity of the
points mr + Oi (where n = ± l, ±2, . . . ,) the relative
error can be quite high although the absolute error is small.
5. The figures cited as the maximum relative errors are
those encountered in a sample of 2500 random arguments
within the respective ranges. See the appropriate
section in Appendix F for a description of the behnvioi
of errors when the argument is near a singularity or a
zero of the function.
6. The maximum relative error cited for the GSIN function
is based upon a set of 2000 random arguments within the
range, !n the immediate proximity of the points
n w + Oi (where n = ± I , ± 2, . , , ,) the relative error
can be quite high although the absolute error is small.
7 , The sample arguments were tangents of numbers
uniformly distributed between - w/2 and + w/2.
1, The answer given is the principal value, i.e. ,
the one whose imaginary part lies between
- w and + tr ,
9, Floating-point overflow can occur.
Section 3: Mathematical Subprogram Descriptions 19
J Table 3. Summary of indirectly referenced mathematical subprograms
2
3
4
5
6
7
8
10
Function
Entry
Name
Definition
Argument(s)
Function
Value
Returned
Error Condition
Storage
Estimates
Other Subprograms Required
Routine
Name
No.
Type
Hex
Dec
RAISE AN INTEGER
BASE TO AN
INTEGER POWER
CHCBGA
y = i * * I
2
2
i = INTEGER * 4
INTEGER * 4
Base is zero
1B4
436
CHCBG
CHCBGB
y= j* * j
j = INTEGER * 2
INTEGER * 2
Base is zero and exponent is
zero or negative
1B4
436
CHCBG
CHCBGC
y=i**i
2
j = INTEGER * 2
i = INTEGER * 4
INTEGER * 4
Base is zero and exponent is
zero or negative
1B4
436
CHCBG
CHCBGD
y = \ ** j
2
i = INTEGER * 4
j = INTEGER * 2
INTEGER * 4
Base is zero and exponent is
zero or negative
1B4
436
CHCBG
RAISE A REAL
BASE TO AN
INTEGER POWER
CHCBHA
y=a**i
2
a =REAL * 4
i = INTEGER * 4
REAL * 4
Base is zero and exponent is
zero or negative
144
324
CHCBH
CHCBHB
y = a * * ]
2
a = REAL * 4
j = INTEGER * 2
REAL * 4
Base is zero and exponent is
zero or negative
144
324
CHCBH
RAISE A DOUBLE
PRECISION BASE TO
AN INTEGER POWER
CHCBIA
y = a * * i
2
a =REAL * 8
= INTEGER * 4
REAL * 8
Base is zero and exponent is
zero or negative
14C
332
CHCBI
CHCBIB
y =a * * j
2
a = REAL * 8
j = INTEGER * 2
REAL * 8
Base is zero and exponent is
zero or negative
14C
332
CHCBI
RAISE A REAL
BASE TO A REAL
POWER
CHCBJA
y = a * * b
2
a - REAL * 4
b = REAL * 4
REAL * 4
Base is zero and exponent is
zero or negative
ICO
448
EXP ALOG
CHCBJ
RAISE AN INTEGER
BASE TO A REAL
POWER
CHCBJB
y = i * * b
2
b = REAL * 4
i = INTEGER * 2
REAL * 4
Base is zero and exponent is
zero or negative
ICO
448
EXP, ALOG
CHCBJ
CHCBJC
y = I**b
2
b = REAL * 4
i = INTEGER * 4
REAL * 4
Base is zero and exponent is
zero or negative
ICO
448
EXP, ALOG
CHCBJ
RAISE A REAL OR
INTEGER BASE TO
A REAL POWER;
BASE AND/OR
EXPONENT
DOUBLE
PRECISION
CHCBKA
y =a * * b
2
a = REAL * 8
b = REAL *8
REAL * 8
Base is zero and exponent is
zero or negative
230
560
DEXP, DLOG
CHCBK
CHCBKB
y = j* *b
2
b = REAL * 8
i = INTEGER * 2
REAL * 8
Base is zero and exponent is
zero or negative
230
560
DEXD, DLOG
CHCBK
CHCBKC
y = i * * b
2
b = REAL * 8
i = INTEGER *4
REAL * 8
Base is zero and exponent is
zero or negative
230
560
DEXP, DLOG
CHCBK
CHCBKD
y = * * b
2
a =REAL * 4
b = REAL * 8
REAL * 8
See Note.
Base is zero and exponent is
zero or negative
230
560
DEXP, DLOG
CHCBK
CHCBKE
y =a * * b
2
a =REAL * 8
b = REAL * 4
REAL * 8
Base is zero and exponent is
zero or negative
230
560
DEXP, DLOG
CHCBK
RAISE A COMPLEX
BASE TO AN
INTEGER POWER
CHCBMA
y = a * * 1
2
a = COMPLEX * 16
i = INTEGER * 4
COMPLEX * 16
Base is zero and exponent Is
zero or negative
274
628
CHCBM
CHCBMB
y =a * * j
2
a = COMPLEX * 16
i = INTEGER * 2
COMPLEX * 16
Base is zero and exponent is
zero or negative
274
628
CHCBM
CHCBCA
y =a * * i
2
a = COMPLEX * 8
i - INTEGER * 4
COMPLEX * 8
Base is zero and exponent Is
zero or negative
24C
588
CHCBC
CHCBCB
y =a * * j
2
a = COMPLEX * 8
i = INTEGER * 2
COMPLEX * 8
Base is zero and exponent Is
zero or negative
24C
588
CHCBC
PRODUCE ERROR
MESSAGE AND
TERMINATE
EXECUTION
CHCBZA
E8
232
As required by use of the
EXIT macro instruction
CHCBZ
NOTE: The REAL*8 function value returned by CHCBKD
given as an argument.
s not more accurate than the REAL*4 base
Section 3: Mathematical Subprogram Descriptions 21
ENTRY NAME : In Table 2, this column shows
the user entry name. In Table 3 , this
column shows the system entry name.
409 6 bytes. The routines are link edited,
and their CSECTs are combined,)
DEFINITION : This column gives a mathemati-
cal equation that represents the computa-
tion, (It is not meant to represent the
way the subprogram is called.) An alterna-
tive equation is given when there is anoth-
er way of representing the computation in
mathematical notation. For example, the
square root can be represented as either
Vx" or x
X-
OTHER SUBPROGRAMS REQUIRED : Many mathemat-
ical subprograms require other mathematical
subprograms to perform their function. The
entry names of the other subprograms are
listed in this column. (This column does
not include CHCBZA, which is called by all
mathematical subprograms where error exit
is possible.)
ARGUMENT (S) : These columns describe the
values (s) for which the function value is
to be computed.
• Argument Number — The number of argu-
ments Cone or two) that the user must
supply.
• Argument Type — The type and length of
the argument. Integer , real, and com-
plex represent the type of number? the
notations *4, *8, and *16 represent the
length, in bytes, of the argument.
Note : In FORTRAN IV f a real argument is a
REAL*** argument, and a double-precision
argument is a REAL*8 argument- A single-
precision complex argument is a COMPLEX* 8
argument, and a double-precision complex
argument is a COMPLEX* 16 argument.
• Argument Range — The valid range for
an argument. (See the Error Condition
and Error Message column descriptions
below. )
FUNCTION VALUE RETURNED:
This column
describes the function value returned by
the subprogram; the notation is the same as
that used for the argument type.
ERROR CONDITION : This column shows the
argument range not allowed when using the
subprogram. If the argument is within this
range, the subprogram will call CHCBZ .
(See the description of CHCBZ, at the end
of this section.)
STORAGE ESTIMATES : This column shows the
approximate number of bytes required for
each mathematical routine: the approximate
total size of each routine's CSECT and
PSECT. (FORTRAN IV mathematical routines
each contain one public, read-only, reent-
erable CSECT and one PSECT. The length of
each of the control sections is less than
ROUTINE NAME : The name of the routine con-
taining the subprogram.
ACCURACY FIGURES (TABLE 2 ONLY) : These
columns give accuracy figures for one or
more representative segments within the
valid argument range. The accuracy figures
are based upon the assumption that the
arguments are perfect (that is, without
error and, therefore, having no error-
propagation effect on the answers) . The
only errors in the answers are those intro-
duced by the subprograms. Information
given in the accuracy-figures columns is :
• Argument Range — This column gives the
argument range used to obtain the
accuracy figures. For each function,
accuracy figures are given more repre-
sentative segments within the valid
argument range. These figures are the
most meaningful to the function and
range under consideration.
The maximum relative error and standard
deviation of the relative error are
generally useful and revealing statis-
tics. However, they are useless for
the range of a function where its value
becomes 0, because the slightest error
in the argument can cause an unpredict-
able fluctuation in the magnitude of
the answer. When a small argument
error would have this effect, the maxi-
mum absolute error and standard devia-
tion of the absolute error are given
for the range. For example, absolute
error is given for sin (x) for values
of x near * .
• Sample — This column indicates the
type of sample used for the accuracy
figures; the type depends upon the
function and range under consideration.
The statistics may be based either upon
an exponentially (E) distributed argu-
ment sample or a uniformly (U) distrib-
uted argument sample.
Section 3: Mathematical Subprogram Descriptions 23
* Statistical results:
M(e)=Max
*>= Vis
f(x)-g(x)
f(x)
f(x.)- S (x.)
f(x.)
M(E)=Max | f(x)-g(x)
'(=)=\n
l V
V
f(xj)— g(xj)
Maximum relative error
produced during testing
Standard deviation (root-
mean-square) of the rela-
tive error
Maximum absolute error
produced during testing
2 Standard deviation (root-
mean-square) of the ab-
solute error,
I Table 5- Exponentiation with real or
double- precis ion base and inte-
ger exponent
Base (A)
Exponent (J)
J>o
J = o
r j < o
A>0
Compute
function value
Function
value = 1
Compute
function value
A =
Function
value =
Error message
Error message
A<0
Compute
function value
Function
value = 1
Compute
function value
In the formulas for the standard devia-
tion 9 N represents the total number of
arguments in the sample; i is a subscripr
that varies from 1 to N.
Test ranges, where they do not cover the
entire legal range of a subroutine, were
selected so that users may infer from the
accuracy figures presented the trend of
errors as an argument moves away from the
principal range. The accuracy of the an-
swer deteriorates substantially as the
argument approaches the limit of the per-
mitted range in several of the subroutines.
This is particularly true for trigonometric
functions. However, an error generated by
any of these subroutines is, at worst, com-
parable in order of magnitude to the effect
c-. the inherent rounding error of the
argument.
Tables 4, 5, and 6 show how the values
of the base and exponent affect the
exponentiation subprograms*
Table 7 shows the system entry names and
compiler- generated entry names of indirect-
reference mathematical routines.
( Table 4. Exponentiation with integer base
and exponent
| Table 6. Exponentiation with real or
double-precision base and
exponent
Base (I)
Exponent (J)
J >
J =
J <
I > 1
Compute the
function value
Function
value - 1
Function
value -.- i.
I = 1
Compute the
function value
Function
value = 1
value ■■■ 1
I =
Function
value =
Error Message
241
Err,,r Messag*
I = -1
Compute the
function value
Function
value = 1
If J is an odd
number, function
value - -1
If J is an even
number, function
value - 1
I < -1
Compute the
function value
Function
value = 1
Function
value =
Base (A)
Exponent (B)
B >0
B =
B<
A>0
Compute
function value
Function
value = 1
Compute
function value
A =
Function
value =
Error message
Error message
A <
Error Message
Function
value = 1
Error message
Table
r~
7. Entry names of indirect-
reference mathematical routines
r T
1 Compiler-
| Generated
Routine
System |
| Entry Name
| FCXPJ
Name
CHCBC
Entry Name |
CHCBCA |
| FCXPI
CHCBC
CHCBCB |
| F1XPI
| CHCBG
CHCBGA I
| FJXPJ
CHCBG
CHCBGB |
| FJXPI
| CHCBG
CHCBGC |
| FIXPJ
CHCBG
CHCBGD |
| FRXPI
1 CHCBH
CHCBHA |
1 FRXPJ
CHCBH
CHCBHB |
| FDXPI
CHCBI
CHCBIA J
1 FDXPJ
| CHCBI
CHCBIB |
| FRXPR
CHCBJ
CHCBJA |
1 FJXPR
1 CHCBJ
CHCBJB |
| FIXPR
CHCBJ
CHCBJC 1
1 * FDXPD
| CHCBK
CHCBKA |
| FJXPD
CHCBK
CHCBKB |
1 . FIXPD
| CHCBK
CHCBKC |
| FRXPD
CHCBK
CHCBKD |
| FDXPR
| CHCBK
CHCBKE I
| FCDXI
CHCBM
CHCBMA I
| FCDXJ
1 CHCBM
CHCBMB |
CHCBZ
CHCBZA* |
L J
L J
L. 1
1* Called by mathematical
library routines |
24
Table 8 describes the macro instructions
used by mathematical library routines. The
mathematical library makes extensive use of
macro instructions for code similarity,
programming efficiency, and ease of inter-
face modification.
CHCBZ — Error Processor
Entry Parameters ; CHCBZ expects general
register 1 to point to a two-word parameter
list. The first word is the address of a
one-byte code indicating the error condi-
tion; codes and meanings are:
the name, CHCBZ does a trace back to
obtain the address of the call to the math-
ematical routine in the user's program.
CHCBZ issues a standard message, inserting
the supplied values and issues the EXIT
macro instruction. The message is:
CHCBZ100 TERMINATED: ARGUMENT $1 FOR
$2 AT $3
where $1 represents the code
$2 represents the name
$3 represents the trace-
back address
- zero
1 - not positive
2 - exceeds limit
3 - negative
4 - absolute value exceeds limit
5 - zero or negative
6 - close to singularity
The second word is a pointer to the name of
the mathematical routine in which the error
occurred, with a one-byte length preceding
Entry Poin t: CHCBZ A
Operation : CHCBZ adds the error code to
the standard system message and issues the
EXIT macro instruction. The standard sys-
tem message is:
OUT OF RANGE PARAMETER GIVEN A FORTRAN
IV SUPPLIED SUBPROGRAM. ERROR CODE IS
nnn.
Table 8. Mathematical library macro instruction summary
Purpose
r r
| Macro Name |
H-
t—
CEKZA
Generates the V-Con/R-Con pair required when other TSS/360 subrou-
tines must be called.
H
CEKZD
CEKTl
CEKT2
CEKT3
Generates the code following a SAVE macro instruction, to add to the
backward and forward chains and load register 13 with the current
PSECT location.
Generates the CSECT label, register EQUs, ENTRY statement, and other
code duplicated at the beginning of every subroutine.
Used for error exits.
Generates PSECT label and save area.
CEKT4
+-
|.
Generates code similar to that of CEKTl, for additional entry
points.
CEKT5 I Similar to CEKT3, but used for PSECT items required for additional
entry points.
_. + _ _ . _
CEKT6 | Used to reload register 13, stop the forward chain, and return.
™t
Section 3: Mathematical Subprogram Descriptions 25
SECTION 4
SERVICE SUBPROGRAM DESCRIPTIONS
This section gives the following infor-
mation on each service subprogram:
* Name
* Name of containing routine
* Function
* Attributes
* Entry
* Routines called
* Error checks
* Data references
* Operation
GENERAL INFORMATION
The three service routines are:
* CHCBE — • interrupt and machine indica-
tor subprograms
* CHCIV — DUMP and PDUMP subprograms
* CHCIW -- EXIT, STOP, and PAUSE
subprograms
CHCBE resides in the mathematical
library; CHCIV and CHCIW reside in the I/O
library.
Like other FORTRAN IV library routines,
the service routines use type-I linkage and
are nonprivileged, reenterable, and closed.
Service subprograms that are user-
referenced have user entry names. No ser-
vice subprograms have compiler-generated
entry names.
Information on entry parameters is con-
tained in the individual routine summaries.
Service subprograms do not pass exit param-
eters, except for the calls to CHCBZ (math-
ematical library error -handling routine) by
the SLITE and SLITET subprograms . CSee
subprogram s ummaries - )
SUBPROGRAM SUMMARIES
Table 9 briefly describes all service
subprograms except the Exit Routine subpro-
grams CHCIWU and CHCIW5, which are
referenced by the I/O library routines.
The following summaries should be read
in conjunction with flowcharts BD, BE f IV,
and IW.
CHCBE — Interrupt and Machine Indicator
Routine
Subprograms :
CHCBE2 — Specification Exception Handler
CHCBE3 — Exponent Overflow Handler
CHCBE4 — Exponent Underflow Handler
CHCBE5 — Divide Check Handler
OVERFL — Exponent Overflow and Underflow
Tester
DVCHK — Divide Check Tester
SLITE — Sense Light Handler
SLITET « Sense Light Tester
Function : To simulate certain machine
indicators, and to mask off or process
interruptions caused by these exceptions:
• Fixed point overflow
• Fixed point divide
• Exponent overflow
• Exponent underflow
• Significance
• Floating point divide
(See Chart BE.)
Entry Parameters : CHCBE2 through CHCBE5
have no entry parameters. OVERFL, DVCHK,
and SLITE each receive one parameter, the
address of an integer variable, A pointer
to this address is passed in register 1.
For SLITET, register 1 points to two
addresses, the address of the integer vari-
able indicating the sense light number and
the address of the integer variable to
receive the result code.
Routines Called :
(CHCBZ)
CHCBZ
Error Processor
Error Checks : If entered at either SLITE
or SLITET, this routine tests the integer
variable and notes an error condition if it
is any value but to 1 for SLITE, or 1 to
26
Table 9. Summary of service subprogram characteristics
1 " 2 3
Function
Entry Name
Error
Condition
Storage
Estimates
Module
Name
HEX
DEC
Pseuoo sense light
subprograms
Turn all sense lights
off or one sense
light on
SLITE
Argument
other than
0, 1, 2, 3, 4
324
804
CHCBE
Test a sense light
or record Its status
SL1TET
Argument
other than
1,2,3,4
CHCBE
Overflow and
underflow
subprogram
Test and record
status of exponent
overflow and
underflow indicators
OVERFL
CHCBD
Divide check
subprogram
Test and record
status of divide
check indicator
DVCHK
CHCBD
Exception
processing
subprograms
Process
arithmetic exceptions
CHCBE3 (EXPONENT OVERFLOW)
CHCBE4 (EXPONENT UNDERFLOW)
CHCBE5 (DIVIDE CHECK)
CHCBE
Process specification
exceptions
CHCBE2 (SPECIFICATION)
CHCBE
Exit subprogram
Terminate execution
EXIT
CHCIW2 (STOP)
CHC1W3 (PAUSE)
1AC
428
CHC1W
Dump subprogram
Dump specified
storage area
with or without
termination
DUMP, PDUMP
48
168
CHCIV
t* for SLITET. If there is an error condi-
tion, CHCBE calls CHCBZA, passing the
appropriate error code as a parameter,
CHCBZA terminates the task after printing
the error code. If entered at CHCBE2 f
CHCBE checks for an interruption that
should not occur and allows the standard
system action for such an interruption.
Operation : At the beginning of all FORTRAN
main programs, the compiler generates code
that calls CHCBD1, the exception processing
enabler and I/O initialization routine
included in the I/O library.
The system interruption handler, CHCBDl f
calls one of the CHCBE2 through CHCBE5 sub-
programs whenever an exception occurs.
CHCBE2 passes control to CHCBE1 to con-
trol fixup of a specification exception.
CHCBE2 checks the interruption address
and the instruction that caused the inter-
ruption to determine whether the interrup-
tion was due to a condition that should not
occur. If so, the standard system action
is invoked. If not, CHCBE1 interprets the
instruction causing the interruption and
executes it as though the alignment restr-
iction was removed. The save area referred
to by register at entry is changed to
reflect instruction execution.
CHCBE3 sets to one the flag which CHCBD6
tests if an "OVERFL (j>* statement occurs
in the user program.
CHCBE4 does the same as CHCBD3, except
that the flag is set to three.
Section 4: Service Subprogram Descriptions 27
CHCBE5 sets to one the flag which CHCBD7
tests if a "DVCHK (j)" statement occurs in
the user program.
OVERFL puts the value of the overflow-
underflow indicator into the user- specif ied
variable, resets the indicator, and returns
control to the user.
DVCHK puts the value of the divide check
indicator into the user-specified variable,
resets the indicator, and returns control
to the user,
SLITE tests the user-specified variable
for a value of zero. If it is zero, all
four indicators are set to zero and control
is returned to the user. If the variable
is not zero, it is tested for values of
one, two, three, and four. If it is one of
these values, the corresponding indicator
is set to one and control is returned ro
the user- If the variable is not one of
these values, SLITE calls CHCBZA and then
returns control to the user.
SLITET tests the first user-specified
variable to see that it is within the prop-
er range. If the variable is in error,
CHCBD9 calls CHCBZA and returns control to
the user. If the variable is from one to
four, the corresponding indicator is
tested. If the indicator is zero, SLITET
sets the second user-specified variable to
two and returns control to the user. If
the indicator is one, SLITET sets the
second variable to one, resets the indica-
tor to zero, and returns control to the
user.
CHCIV
Dump Routine
This routine causes the user- indicated
limits of storage to be dumped in the for-
mat desired, with or without program ter-
mination- The user calls this module ijy
specifying either of the statements CALL
DUMP (A ±r B ±i F ±r . .., A n , B nr F n ) or CALL
PDUMP (A ± , B ± , F ±t . .., A n , B n , F n > in his
source program. The variable data names, A
and B, indicate the limits of storage to be
dumped and the integer, F, indicates the
dump format desired. (See Chart IV.)
For sample printouts by the DUMP and
PDUMP subprograms, see FORTRAN IV Library
Subprograms , GC28-2026, "Appendix D: DUMP
and PDUMP Sample Storage Printouts."
Entry ; There are two entry points: CHCIV1
and CHCIV2, for DUMP and PDUMP, respective-
ly. Standard type-I linkage is used with
register 1 pointing to the first address
constant in the parameter list. The param-
eter list is variable- length and has the
following format :
h
Word 1
Word 2
Word 3
Word 4
First
Group
Word 5
Second
Word 6
Group
Word 7
L
Word size of this parameter
list, minus one.
Address constant pointing
to the starting location
symbol for the dump.
Address constant pointing
to the end location symbol
for the dump.
Address constant pointing
to the FORMAT code word.
When FORMAT code is zero,
the FORMAT code word is
divided into two parts:
first half-length of end
location symbol minus one
second half - 0.
Same as words 2, 3, and 4
respectively.
The entries in the parameter list
(excluding the first word) are groups of
three words. There may be more than one
group depending on the number of different
areas of virtual storage the user wants
dumped. The first two words of a group may
be in either order; but the dump is always
taken in ascending order. The third word
of a group, the format code word, repre-
sents the type of data conversion to be
performed, as follows:
hexadecimal
real* 4
1
logical*l
6
real*8
2
logical***
7
compl ex* 8
3
integer*2
8
complex* 16
4
integer* 4
9
literal
The format code is the default value.
Routines Called :
• Integer Output Conversion (CHCIH)
• Complex Output Conversion (CHCIN)
• Logical Output Conversion (CHCIR)
• General Output Conversion (CHCIT)
• Exit (CHCIW)
• GATE facilities (GATWR macro
instruction)
28
Error Checks : CHCIV makes no error checks;
all error checking is done by the called
data conversion routines.
Operation : CHCIV scans the parameter list
to locate a parameter group. The specified
dump area is then formatted and written on
SYS0UT using the GATE macro instruction
facility. After the parameter list has
been completely processed, control returns
to the user program, if PDUMP was speci-
fied. If DUMP was specified, control
passes to the exit routine (CHCIW) for pro-
gram termination.
CHCIV determines the presence of the
third word in a parameter group by testing
the value of the word in storage to which
it refers. If the value is 9 or less, it
is assumed to be the format code. If the
value is greater than 9 (or if it is nega-
tive) , it is assumed that the third word
was omitted and a format code of zero
(hexadecimal) is taken as a default. If
the format code is either or 9 (hexade-
cimal or literal) , the conversion format-
ting is done within this routine; other-
wise, the appropriate output conversion
routine is called.
CHCIW3 Execution of a PAUSE statement
CHCIW** Exit requested by other I/O
library routines
CHCIW5 Exit requested by a user subpro-
gram error
Standard type-1 linkage is used with all
entries except CHCIW**, which uses type-4
restricted linkage. In the case of STOP
and PAUSE statements, the address of a pa-
rameter list is passed in register 1. The
parameter list is fixed- length and has the
following format:
r T . ,
| Word 1 | Address of byte containing the
I length of the integer or message
| (n or 'message*). A length of
| zero indicates that no integer or
(message was defined by the user.
j. + _ i
(Word 2 [Address of the integer or message,
j if any.
i- . — x . . . . i
There is no parameter list for entry to
CHCIW1, CHCIW4, or CHCIW5.
CHCIW — Exit Routine
This routine performs user-program exit
functions whenever any of the following
FORTRAN statements are executed:
CALL EXIT
STOP n
PAUSE n
PAUSE •message*
In the case of a STOP or PAUSE state-
ment, this module causes a one- through
five-digit integer (n) or a message ('mes-
sage') to be produced at the user's SYSOUT.
With the exception of the PAUSE statement,
CHCIW causes all DCBs in use in the DCB
table to be closed. (See Chart IW.)
Entry : The five entry points and their
associated causes of entry are as follows:
CHCIW1 Execution of a CALL EXIT
statement
CHCIW2 Execution of a STOP statement
Routines Called :
• DCB Maintenance (CHCIB)
• I/O Control (CHCIC)
• Task Supervisor Routines (CZAFQ1) ,
entered by use of a supervisor call
| (SVC)
• I/O Communication (CHCIY)
• PRMPT (CZATJ1)
Operation : If this routine is entered via
CHCIW4 or CHCIW5, a message is issued to
| the user indicating the reason it was
called. If this routine is entered via
CHCIW1, CHCIW2, or CHCIW3, a message is
issued (including a user-supplied message,
if any) to the user by the PRMPT macro
instruction facility (CZATJ1) indicating
the reason it was called. In the case of
PAUSE, return is then made to the user pro-
gram. If entry to this routine was not
caused by a PAUSE statement, the DCB table,
maintained by CHCIB, is searched and each
open DCB is closed. The DCB table pointers
are then cleared, and exit is made to the
system exit routine.
Section 4: Service Subprogram Descriptions 29
SECTION 5:
I/O ROUTINE DESCR IPTIONS
This section gives the following infor-
mation on each I/O library routine, exclud-
ing the service routines, which are
described in Section 4s
• Name
• Function
• Attributes
• Entry
» Routines called (where applicable)
• Error checks (where applicable)
• Data references (where applicable)
• Operation
The routines are described in the alpha-
betical order of their names .
ROUTINE SUMMARIES
CHCIA
I/O Initialization
This routine is the initial FORTRAN I/O
library interface with the user. It
manages the disposition of each I/O request
by setting switches about formatted and
unformatted I/O for the information of
other I/O routines, by allocating a buffer
area for output requests, and by obtaining
a logical record for input requests.
Every I/O request in the FORTRAN user
program causes the compiler to generate one
call to CHCIA. On this call, if there is
no list, CHCIA supervises the complete
execution of an I/O request. If the I/O
request is a READ, WRITE, PRINT, or PUNCH
with list, CHCIA simply prepares the I/O
library for compiler-generated calls to
List Item Processor (CHCIE) and List Ter-
mination (CHCIU) .
GENERAL INFORMATION
Certain information is common to most
I/O routines. This information includes:
• Names
• Attributes
• Format of parameters passed to data
conversion routines
Attributes
All FORTRAN I/O routines are nonprivi-
leged, reenterable, and closed. CHCIA,
CHCIE, CHCIU, CHCIW1, CHCIW2, CHCIW3,
CHCIW5, CHCIV1, CHCIV2, and CHCBDl are
entered by standard type-1 linkage with the
address of a parameter list (where applic-
able) in register 1? exit is a return to
the calling routine. All of the other I/O
routines are entered by restricted type- 4
linkage.
Work Areas and Register Save Areas
The FORTRAN I/O routine has one common
PSECT — CHCRWW — which contains all
necessary work and storage areas and a 19-
word register save area. All parameters
used by the data conversion routines are
stored in this PSECT.
Entry : The entry point is CHCIA1. The pa-
rameter list is variable- length and has
this formats
H _ -[ 1
| Word 2 | Address of a control byte in I/O ]
Control (CHCIC) indicating type of
operation: READ, WRITE, PRINT,
PUNCH, REWIND, BACKSPACE, ENDFILE.
Word 3 (Address of a control byte in I/O
Control (CHCIC) indicating whether
a list was present in the I/O
statement and whether any of the
following parameters in this list
are present.
Word 1
h
h
Word
(Op-
tion-
al)
Address of a fullword containing
the user-specified data set
reference number.
H
Word 4 | Address of a FORMAT string or
(Op- |NAMELIST table. This parameter
tion- J is included only if the I/O source
al) I statement has an associated FORMAT
or NAMELIST source statement.
Word 5 [Address of an error exit.
(Op- | Included only if the I/O source
tion- | statement has an ERR operand.
al)
Address of an end-of-file exit.
Included only if the I/O source
statement has an END operand.
30
If any optional parameter is missing,
any parameters following it are moved up in
the list and the list is shortened. For
example, if there is no FORMAT or NAMELIST
address and no error exit address, word *t
of the parameter list would be the end- of -
file exit address.
Routines Called :
• DCB Maintenance (CHCIB)
• I/O Control (CHCIC)
• FORMAT Processor (CHCIF)
• NAMELIST Processor (CHCID)
Entry : The entry point is CHCIB1. CHCIA
stores the address of the user-supplied
data set reference number and the data set
reference number itself if present in the
I/O common PSECT.
Routines Called :
• Data management routines used to search
for and read JFCBs (CZAEB)
• Data management routines used to allo-
cate storage for DCB construction
(CZCGA)
• Exit (CHCIW)
• Exit (CHCIW)
• PRMPT (CZATJ1)
Error Checks : If the user-specified data
set reference number is negative, an error
message is issued by the PRMPT macro
instruction facility (CZATJ1), and CHCIW is
entered to terminate the user program.
Operation : Upon entry, the FORTRAN I/O
PSECT switches are cleared, control parame-
ters are set, and CHCIB is called to find a
DCB corresponding to a user-specified data
set reference number (or to create a DCB if
one is not found) . CHCIC is then called to
complete the I/O request.
If the operation is a REWIND, BACKSPACE,
or ENDFILE, return is made to the user. If
the operation is READ or WRITE according to
a NAMELIST, CHCID is called before return-
ing to the user.
If the operation is a WRITE with no list
and no format, a dummy record containing
two bytes of zeros is written with a call
to CHCIC before returning to the user. If
a FORMAT statement is involved, CHCIF is
called. If the operation is a WRITE with
no list and CHCIF is called for construc-
tion of a record anyway, CHCIC is called to
write the record before returning to the
user? otherwise, return is made immediately
after calling CHCIF. If a FORMAT statement
is not involved, return is made to the
user.
CHCIB — DCB Maintenance
This routine finds or initializes the
data control block (DCB) that contains a
description of the data to be transmitted
by a user-specified I/O operation. If an
appropriate DCB is not found, this routine
allocates the necessary space in the DCB
table and constructs a new DCB, including
within it information about the data to be
transmitted that the user defined in his
DDEF command. (See Chart IB.)
• PRMPT (CZATJ1)
Error Checks : If the user-specified data
set reference number exceeds 99, an error
message is issued by the PRMPT macro
instruction facility (CZATJ1) , and CHCIW is
entered to terminate the user program. No
alternate data set reference number is
established.
If a discrepancy exists in the user DDEF
command between permissible RECFM, KEYLEN,
and DSORG values, an error message is
issued by the PRMPT macro instruction faci-
lity (CZATJ1), and CHCIN is called to ter-
minate processing. A description of the
assumptions FORTRAN I/O makes an initializ-
ing associated DCBs is contained in Appen-
dix A.
Format and Content of the DCB Prefix : The
DCB Prefix is used by the FORTRAN I/O rou-
tines, in conjunction with the DCB, when
performing any type of I/O operation. The
DCB Prefix, created by the DCB Maintenance
routine (CHCIB) , is eight words long and
always immediately precedes the DCB itself.
(See Table 10.)
Operation : This routine is called to per-
form one of three functions: to supply the
address of an initialized DCB corresponding
to a specified data set reference number;
to locate the first open DCB in the DCB
table used by the I/O library routines; or
to obtain a GATE DCB. The first function
is performed by searching the table of DCB
entries. If the appropriate DCB is found,
the address of that DCB is returned. If
the search fails, the data management rou-
tine CZAEB is called to determine whether
the user has issued a DDEF command (which
causes the creation of a Job File Control
Block ~ JFCB) for the specified data set
reference number. If CZAEB finds a JFCB
that corresponds, space is reserved in the
DCB table, a DCB is constructed, and the
new DCB is "chained" into the DCB table.
If CZAEB cannot locate a corresponding JFCB
(meaning that the user did not supply a
Section 5: I/O Routine Descriptions 31
Table
r-
Word
10,
Format and content of DCB prefix
I
Word
I-
Address of the starting location
in the buffer area for the current
logical record.
Word 3
Address of the current location in
the buffer area for the current
logical record.
Address of the end location in the
buffer area for the current logic-
al record.
Byte Is Current operation (READ,
WRITE, etc.)
Byte 2: Control flags (FORMAT,
NAMELIST, List, ERR exit,
END exit)
Word 4
.-+
Word 5
h
Word 6
k
Word
\~
Word
\
Word
~\
Byte 3s Control flags (Span, GATE,
recent READ, END or ERR
encountered)
Byte 4: Previous operation (byte 1
from last call on CHCIC
with this DCB)
Address of current DECB, if
required (BSAM) .
H
User-specified data set reference
number, plus one.
Address of the next DCB.
Address of last DCB associated
with this data set reference
number.
9 j DCB begins here.
1
— j
DDEF command) , a GATE DCB is constructed
and initialized*
The GATE DCB causes SYSIN and SYSOUT to
be treated as the input origin and output
destination, respectively.
CHCIB performs its second function,
locating the first open DCB in the DCB
table, when a negative data set reference
number has been specified by CHCIW at
final close of the file. If an open DCB
is not found in the DCB table, normal
return is made to the calling routine.
CHCIB performs its third function,
obtaining a GATE DCB, when the address of
the data set reference number is zero. In
this case, the I/O request is assumed to
be for the first data set reference number
that this routine defines by a GATE DCB.
Therefore, as before, SYSIN and SYSOUT are
treated as the input origin and output
destination, respectively. Note that the
GATE DCB is a dummy DCB (with a prefix) ,
used to allow other I/O routines to treat
the case of GATE I/O the same as the case
of data sets defined by a DDEF command •
CHCIC — I/O Control
This routine fulfills I/O requests made
through other I/O library routines by
using the data management macro instruc-
tion facilities of TSS/360. The particu-
lar data management facilities to be used
are determined both by the type of I/O
statement issued in the user program, and
by related DDEF commands, if any, defining
such things as the type of records being
transferred and the manner in which they
should be processed.
The following list identifies the more
significant macro instructions used by
CHCIC for each of the FORTRAN I/O state-
ments. Other macros used in conjunction
with those listed below (e.g., OPEN and
CLOSE) are identified in the "Operation"
section of this routine description. (See
Chart IC.)
FORTRAN I/O
Statement
READ
WRITE
REWIND
BACKSPACE
ENDFILE
CHCIC Function
Obtains a logical record
from a user- specif ied input
source by using the READ,
GATRD, or GET macro
instruction.
Initializes the writing of a
logical record by establish-
ing pointers to the output
buffer area. Subsequent
output processing is per-
formed by using the WRITE,
GATWR, or PUT macro
instruction.
Repositions the user-
specified volume of one or
more data sets to the first
record of the first data set
by using the POINT or SETL
macro instruction.
Repositions the user-
specified data set to the
previous logical record by
using the NOTE, POINT, SETL,
and BSP macro instructions.
Defines the end of the user-
specified data set by using
the WRITE and STOW macro
instructions .
32
Entry : The entry point is CHCIC1.
Routines Called :
• DCB Maintenance (CHCIB)
• Exit (CHCIW)
• Data management routines to perform
I/O functions as determined by the
macro instruction issued.
• PRMPT (CZATJ1)
Error Checks : If the I/O operations per-
formed by data management cause either a
SYNAD or EODAD exit, and if the user pro-
vides an ERR or END return point, CHCIC
locates the adcons for these return points
in the work area CHCRWW, and locates the
register save area for the user's program
registers. Return is then made to the ERR
or END return point rather than to the
calling I/O routine.
If the user does not provide return
points (or if the operation is other than
a READ statement) f an error message is
issued and the program is terminated.
If an invalid character is encountered
in hexadecimal input from a GATRD opera-
tion performed for an unformatted READ
statement, an error message is issued and
the erroneous character is treated as the
termination of the hexadecimal input.
Processing then continues.
In addition to the above error checks,
error messages are issued (PRMPT macro
instruction) and the user program is ter-
minated by CHCIW for any of the following
reasons :
1. The record is not format- V for unfor-
matted READ statement.
2. Error return code received from the
use of the FIND or STOW macro
instruction for a member in a VPAM
data set.
3 . Invalid sequence of I/O operations
for a user-specified data set
reference number. The invalid
sequences are: READ preceded by END
FILE; END FILE preceded by READ? and
READ preceded by WRITE (except when
using GATE I/O) .
Operation : In fulfilling the I/O request,
this module ensures that the DCB asso-
ciated with the DDEF command is properly
opened by examining the access qualifier
(for example, read-only) of the data set.
If the data set is read-only, the OPEN
macro instruction is issued for INPUT.
Otherwise, the OPEN macro instruction is
issued for UPDAT, when the Virtual Access
Methods (VAM) are specified, or INOUT,
when the Basic Sequential Access Method
(BSAM) is specified. When a data set is
opened for UPDAT or INOUT, both reading
and writing may be performed. However, if
an attempt is made to write on a read-only
data set that is opened for INPUT, the
user program is abnormally terminated.
When the user does not issue a DDEF com-
mand for a given I/O request, CHCIC uses
the GATE macro instruction facility to
perform reading and writing. This facili-
ty does not require issuance of an asso-
ciated OPEN macro instruction. CHCIC
first checks whether the I/O operation
requested is compatible with the previous
operation. If it is not, an error message
is issued and the user program is ter-
minated. If it is compatible, processing
continues according to the particular type
of I/O request involved (that is, READ,
WRITE, WRITE Initialize, REWIND, BACKS-
PACE, and ENDFILE).
If a READ is issued, the data set is
opened, if necessary, and the pointers in
the DCB prefix are set to the beginning
and end of the data. CHCIC then checks
which access method is being used. If VAM
is used, a GET macro instruction is
issued. If the GATE macro instruction
facility is used, a GATRD macro instruc-
tion is issued. If BSAM is used, the
record format is examined to determine
whether the records are blocked. If they
are not blocked or if the previous record
was the last record of the block, a READ
is issued; otherwise, the next logical
record within the current block is used.
In any case, after the data is located,
the beginning and end are determined and
the pointers are saved in the DCB prefix.
Return is then made to the calling
routine.
If a WRITE Initialize is issued for an
open data set for which the last operation
was an END FILE, the data set is closed
and an end file lockout indicator is set
on in the DCB prefix, indicating to DCB
Maintenance (CHCIB) that this DCB should
be ignored when looking for a DCB for this
data set reference number. As a result of
this indicator being set, any subsequent
WRITE statements referring to the same
data set reference number causes the crea-
tion of a new data set. A call is then
made to CHCIB to obtain a new DCB describ-
ing the new data set.
Whether or not the previous operation
was an END FILE, the DCB is opened, if
necessary. If VAM is used, a SETL macro
instruction is issued to point to the
beginning of the specified data set. In
all cases, the pointers to the output
Section 5: I/O Routine Descriptions 33
buffer area are then set in the DCB prefix
and return is made to the calling routine.
If a WRITE is issued, CHCIC checks the
record format type. For non-format-V rec-
ords, fill characters (padding) are
inserted, if necessary, to the end of the
record. Blanks are used for NAMELIST or
formatted WRITE; binary zeros are used
otherwise. For format-V records, the
record length and any needed spanning bits
are set. The spanning bits are used to
define the relationship between the FOR-
TRAN logical record and the data manage-
ment logical record when the FORTRAN log-
ical record exceeds the length of the
maximum record supported by the access
methods being employed. CSee Appendix A
for more details on spanning bits.)
CHCIC next checks the access method
being used for the write request. For
VAM, a PUT macro instruction is issued.
For the GATE macro instruction facility, a
GTWRC is issued. For BSAM, the record
format is examined to determine whether
the records are blocked. If the records
are not blocked or the block is full, a
WRITE macro instruction is issued. For
blocked records, where the block is not
full, the output area pointers in the DCB
prefix are updated to the next record area
in the block. In all cases, the DCB pre-
fix is re-initialized and return is made
to the calling routine.
If a REWIND is issued, CHCIC searches
the DCB table for all DCBs relating to the
user-specified data set reference number.
For each such DCB found, the end file
lockout indicator within it is turned off*
The POINT macro instruction is then used
to reposition the volume to the first
record of the first data set, if the data
set is BSAM; the SETL macro instruction is
used for VAM. Return is then made to the
calling routine.
If a BACKSPACE is iss
the access method being
VAM, CHCIC checks to see
operation was an ENDFILE
macro instruction is iss
OPEN macro instruction a
instruction pointing to
data set. In either cas
instruction is then issu
previous record.
ued, CHCIC checks
employed. For
whether the last
If so, a CLOSE
ued followed by an
nd a SETL macro
the end of the
e, a SETL macro
ed to point to the
If BSAM is used, CHCIC examines the
record format to determine whether the
records are blocked. If unblocked and if
the last operation was an END FILE, a NOTE
macro instruction is issued to establish
the current location, CLOSE and OPEN macro
instructions are issued, and a POINT macro
instruction is issued to reposition the
data set to the position prior to closing.
A BSP macro instruction is then issued to
backspace. If double buffering is used, a
second BSP is issued. If the last opera-
tion was not an END FILE, only the BSP
macro instruction is issued. If the rec-
ords are blocked, a scan is made backwards
through the block to locate the beginning
of the previous record. In some cases it
may be necessary to issue a BSP macro
instruction and then read the previous
block in order to position to the previous
record. In all cases, it is not possible
to position to a record in a previous data
set if a READ or WRITE was issued after an
END FILE operation. Upon completion of
processing, return is made to the calling
routine.
If an END FILE is issued, it is noted
but otherwise ignored; the actual END FILE
operation is performed only when the next
operation is a WRITE Initialize or a
BACKSPACE.
CHCID
NAMELIST Processor
This routine interacts with CHCIC to
control the I/O for each NAMELIST record
and interacts with the appropriate data
conversion routines to effect the desired
item-by-item conversion. CSee Chart ID.)
Entry : The entry point is CHCID1.
R outines Called :
• I/O Control CCHCIC)
• Complex Input Conversion (CHCIM)
• Complex Output Conversion (CHCIN)
• General Input Conversion (CHCIS)
• General Output Conversion (CHCIT)
• Exit (CHCIW)
• PRMPT (CZATJ1)
Error Checks : There are no error checks
for output. For input, if errors are
detected in the NAMELIST table, a message
is issued via PRMPT, and CHCIW is called
to terminate the user program. Other
error messages are generated for any of
the conditions listed below. In these
cases, processing continues with the next
entry of the input record.
• Name exceeds six characters
• First character of each input record
is not blank
• Subscripts appear on a name that is
not an array name
34
• Incorrect number or range of
subscripts
• Subscripting causes array size to be
exceeded
• Multiple constants or repeated con-
stants appear with a name that is not
an array or a subscripted array name
• Multiple constants or repeated con-
stants exceed the size of an array or
the size of an array portion beginning
at a specified element
• An equal sign or left parenthesis is
not preceded by the variable or array
name for that item
• An invalid character appears in a
repeat constant
• End of a logical record caused an item
to be logically incomplete
• The NAMELIST name is not in the NAME-
LIST table
NAMELIST Table : The address of the NAME-
LIST table generated by the FORTRAN com-
piler is communicated in the call to I/O
Initialization CCHCIA) and then passed to
this routine. The table is made up of
two-word entries, each of which contains
an identifier in the first halfword.
NAMELIST Name Entry:
Bytes 0-1 Identifier (X'OIOO 1 )
2-7 Name (left- justified)
Variable Name Entry:
Bytes 0-1 Identifier (X f 0200 f >
2-7 Name Clef t- justified)
Variable Type and Location Entry:
Bytes 0-1 Identifier (X^OO*)
2 Length and Type (4 bits each)
Length:
Type:
3 Class :
Number of bytes
minus 1
X'Ql 1 Logical
X , 02 i Integer
X f 03 f Real
X , 04 f Complex
Letter A for array ;
otherwise, an S
Array Size Entry:
Bytes 0-1 Identifier (X , 0400 i )
2-3 Not used
4-7 Number of bytes in array
Dimension Product Entry:
Bytes 0-1 Identifier (X'OSOO 1 )
2-3 Not used
4-7 Dimension Product (see
explanation below)
Terminal Entry:
Bytes 0-3
4-7
Zero
Not used
4-7 Storage Location
A dimension is a level of subdivision,
or level of subscripting, within an array.
For example, an array could be a string of
seven thirty-word elements (first dimen-
sion), each subdivided into six five-word
elements (second dimension) , each subdi-
vided, into five one-word elements (third
dimension). An array may have as many as
seven dimensions.
For each dimension there is a corres-
ponding dimension product , which is the
product of: (1) the byte-size of the
array's smallest element, (2) the number
of elements within all lower dimensions
except the first dimension, and (3) the
number of elements within that dimension.
In the example just given, the dimension
product for the third dimension would be
4x6x5, or 120. This dimension product
would be seven times greater if there were
another dimension before the seven- element
dimension. The dimension product for the
first dimension is always the byte-size of
the array's smallest element — this
dimension product is never entered. If
there is only one level of subdivision,
there is no Dimension Product Entry.
Operation : For a WRITE, CHCID calls CHCIT
to write, in the following format, the
data for each variable or array in the
NAMELIST:
For Variables:
variable name = constant
For Ar r a ys :
array name or array element name =
constantl, constant2, ...
In the second case, each constant may
have a repetition notation appended in the
form that is also usable for input.
CHCID also produces header and trailer
records so that the information can be
read with the same NAMELIST.
Section 5: I/O Routine Descriptions 35
For a READ, CHCID reads records until a
correct header record is located. It then
uses a scanning procedure to locate the
input entries. As each is found, the rou-
tine locates the variable or array name in
the NAMELIST Table, determines its size
and type f and converts the input.
Error checking functions performed by
CHCID for a READ operation have already
been described under "Error Checks",
CHCIE — List Item Processor
Every I/O statement in the user's
source program generates one or more calls
to this routine if there is a list asso-
ciated with a READ, WRITE, PRINT, or
PUNCH. A list item may be a simple vari-
able, an array element (a subscripted
variable), or an entire array. If a FOR-
MAT statement is specified, this routine
calls on FORMAT Processor (CHCIF) to con-
trol any necessary conversion. If there
is no FORMAT statement, CHCIE is directly
responsible for filling or emptying the
output or input buffer area. (See Chart
IE.)
Entry : The entry point is CHCIE1. Regis-
ter contains either zeros, if the list
item is a single element, or a number
expressing the array length, in bytes, if
the list item is an entire array. The pa-
rameter list is fixed- length and has the
following format:
Address of a control byte. The
first four bits of the control
byte contain the size of the
element, minus one. The second
four bits contain a flag
indicating the type of item as
follows :
Flaq
Type of Item
01
logical
02
integer
03
real
04
complex
Address of a first (or only) ele-
ment of the list item.
Error Check ; With unformatted input, if a
list item is requested after the logical
record is exhausted, an error message is
transmitted to the user via PRMPT, and
CHCIW is called to terminate the user's
program.
Operation : At entry, register either
contains zeros if the list item is a single
element, or a number expressing the array-
length, in bytes, if the list item is an
entire array. Another parameter indicates
the length of the list item (in the case of
an array, the length of each individual
item in the array) . The transfer of data
between I/O buffers and item storage loca-
tions is performed by CHCIE if there is no
FORMAT statement,- if there is a FORMAT
statement, the transfer of data is accomp-
lished by a call to CHCIF. After the
transfer of information between the I/O
buffer and the storage location specified
for the item, the item length is subtracted
from the array length. A negative or zero
result indicates either that the item was a
single element or that the last element in
the array has been processed? the routine
returns to the caller. If the result is
positive, the storage location of the list
item and the location in the buffer are
each incremented by the item length and
transfer is made to the beginning of the
loop to process the next element of the
array. If, after incrementing the buffer
location, the end of the buffer has been
reached, further processing depends on
whether the operation is for input or
output.
For output, the current buffer is writ-
ten and flagged to indicate that this log-
ical record continues on the next physical
record. A new output buffer is then
initialized.
For input, a check is made to see if the
current physical record is flagged in this
manner, and if it is, the next physical
record is read. If the current physical
record is the last one written as part of
this logical record, an error message is
transmitted to the user and the task is
terminated. The reading and writing of
additional physical records as part of this
logical record is accomplished by a call to
I/O control (CHCIC).
CHCIF — Format Processor
Routines Called :
• Format Processor (CHCIF)
* I/O Control CCHCIC)
* Exit CCHCIW)
• PRMPT CCZATJ1)
This routine scans the FORMAT string,
interacts with CHCIC to control the I/O for
each FORMAT-referenced record, and
interacts with the appropriate data conver-
sion routines to effect the item-by-item
conversion specified by the referencing
FORMAT statement. CHCIF is entered once
for each list item in the I/O request.
(See Chart IF.)
36
Entry : The entry point is CHCIF1. Before
the first entry to CHCIF to process a
reference to a FORMAT statement , CHCIA (I/O
Initialization) stores the address of the
FORMAT string in CHCRWw.
Immediate input or output of logical
records
Initialization of return points to be
used when closing parentheses are
reached
Routines Called:
• I/O Control (CHCIC)
Immediate conversion of input or output
data in accordance with the FORMAT con-
version code
• Exit (CHCIW)
• One of the eleven conversion routines
(CHCIH through CHCIT)
• The FORTRAN compiler routine for trans-
lating FORMAT statements (CEKBF), at
entry point SYSPFMT
• PRMPT (CZATJ1)
Error Checks : Since FORMAT statements may
be dynamically modified, certain error con-
ditions may arise due to the syntax of the
FORMAT string. If there are no syntax
errors, errors could arise due to conver-
sion of the data. In such cases the con-
version routines issue messages describing
the errors before returning. All syntax
error checks produce messages describing
the error.
In the last case, if an item is known to
be available, the appropriate conversion
routine is called and return is made to the
calling routine. If no item is available,
the scan continues.
On entries for subsequent list items in
the I/O request, CHCIF keeps track of its
position in the FORMAT table.
The scan ends when a closing parenthesis
is reached.
CHCIH — Integer Output Conversion
This routine converts a two-byte or
four-byte binary list item to an integer
field in the output buffer, according to
the format In, where n is the integer
field's size. (See Chart IH.)
Processing is terminated upon encounter-
ing invalid control characters in the
string, strings that exceed the maximum, or
too many levels of parentheses. When it is
possible to assume values other than those
specified (as in the case of invalid size
of w or d fields after a control charac-
ter) , processing continues on the current
item after the error message is issued.
Otherwise, the erroneous FORMAT item is
skipped, and processing continues with the
next control character.
Operation : At the first entry to CHCIF for
a given I/O request, CHCIF tests whether
the FORMAT statement has been translated.
If not, CHCIF calls CEKBF at entry point
SYSPFMT. CEKBF enters translations of FOR-
MAT codes into a FORMAT table, as shown in
Table 11.
CHCIF scans the translated FORMAT
string, tests each entry for validity, and
performs these actions:
• Initialization of repetition factors
both inside and outside the FORMAT
parentheses
• Immediate input or output of character
strings or spaces (wH or w or Tw or
• Initialization of scale factors before
FORMAT control units
Entry : The entry point is CHCIH1. The pa-
rameter list is described under "Data Con-
version Routine Parameter Lists," in this
section.
Routine Called : PRMPT (CZATJ1)
Error Checks : If the output buffer area is
too small to contain the integer field, the
field is filled with asterisks and a mes-
sage is issued by PRMPT.
Operation : The contents of the list item
are converted in a work area to packed
decimal, then into zoned decimal. The
sign, if negative, is set in the conven-
tional FORTRAN position. The work area is
moved into the output buffer field from
right to left, to check whether sufficient
space is supplied by the field width.
Blanks are inserted as padding if the buff-
er size is larger than the converted field
size.
CHCII — Real and Integer Input Conversion
This routine converts a data field in an
input buffer to the appropriate type of
list item. An integer field is converted
to a binary list item. A real field is
converted to a single- or double- precis ion,
floating-point list item. The integer
field has the format In, where n is the
field width. The real field has the format
Fw.d, Ew.d, or Dw.d, where w is the field
Section 5: I/O Routine Descriptions 37
Table 11 . Translation of FORMAT codes
FORMAT
CODE
FORMAT TABLE ENTRY
SIZE
(BYTES)
BYTE
BYTE
1
BYTE BYTE BYTE
2 3 4
BYTE
5
H
STRING
LENGTH-2
X'O'
LENGTH
CHARACTER STRING
AS MANY BYTES AS NEEDED (MAX 255)
/
1
XT
X
2
X'2«
REPEAT
COUNT
T
2
X'3'
W
P
2
X r 4*
SCALE
FACTOR
)
2
X'5'
NEST
LEVEL
(
3
X»6'
NEST
LEVEL
REPEAT
COUNT
A
3
X'7'
REPEAT
COUNT
W-l
Z
3
X'S'
REPEAT
COUNT
W-l
L
3 i
X'9»
REPEAT
COUNT
W-l
I
3
X'A»
REPEAT
COUNT
W-l
G
4
X'B'
REPEAT
COUNT
W-l 1 D
F
4 ;
X'O
REPEAT
COUNT
W-l
D
D
4
X'D'
REPEAT
COUNT
W-l
D
E
4
X'E'
REPEAT
COUNT
W-l ID
1
SPECIAL
H
6
X'F'
LENGTH
....
ADDRESS OF CHARACTER
STRING
width and d is the width of the decimal
fraction* (See Chart II.)
Entry ; There are three entry points:
CHCII, CHCIK, and CHCIG.
Routine Called: PRMPT CCZATJ1)
Error Checks s If the FORMAT specification
CF ff E, D f or I> is improperly specified or
the data field is greater than the permis-
sible range, PRMPT is called.
Operation ; The input data field is scanned
from left to right to determine:
1. The value of the sign, if any.
2. Position of the base number's first
digit.
3. Number of digits before the decimal
point, if real.
4. Position of first digit after decimal
point, if real,
5. Number of digits after decimal point,
if real.
6. The letter E or D f if present.
7. Sign preceding exponent, if any.
8. Position of exponent's first digit.
9. Number of exponent digits.
38
During the scan a test is made for in-
valid characters in the numeric fields. In
addition, a test is made to determine if
the integer or real data exceeds the per-
missible maximum. If such errors are
encountered, the scan is terminated and
control is passed to PRMPT to issue an
error message. Otherwise, conversion con-
tinues and a normal return is made.
CHCIJ — Real Output Conversion
This routine converts a single- or
double-precision floating-point list item
to a real field in the output buffer. The
real field has a format of either Ew.d,
Dw.d, or Fw.d, where w is the field width
and d is the size, in digit positions, of
the fractional position. (See Chart IJ.)
Entry : There are two entry points:
and CHCIL1.
CHCU1
sing parenthesis or central comma, CHCIM
issues an error message via PRMPT. No
further action is taken and the list items
remain unchanged. If either or both real
fields contain invalid characters or exceed
the permissible magnitude, CHCII issues an
error message.
Operation : The input buffer is scanned to
find the size of each real data field.
Once it is determined that the user proper-
ly specified these data fields and included
the correct delimiters (parentheses and
central comma) , the first real data field
is passed as a parameter to CHCII for con-
version to a single- or double- precis ion
floating-point list item. When CHCII
returns control to CHCIM, CHCIM passes the
second real data field to CHCII for conver-
sion, completing the complex input
conversion.
Routines Called : Return is made to the
calling routine.
Error Checks : If the output buffer is too
small to contain the real field, the real
field is filled with asterisks.
Operation : A test is made against a table
of powers of ten to determine the required
output exponent, if any, and the item is
divided by the appropriate power of ten.
If there is an exponent, it is moved into
the output buffer from right to left, fol-
lowed by its sign and the letter E or D.
If there is no exponent, the data field,
after processing is complete, is moved into
the output buffer and control is passed to
the calling routine.
CHCIM — Complex Input Conversion
This routine converts a complex data
field in the input buffer to a complex list
item consisting of two real data fields.
Each real field is converted to a single-
or double- precis ion floating-point list
item according to the format Fw.d, Ew.d, or
Dw.d, where w is the real field width and d
is the width of the decimal fraction. (See
Chart IM. >
Entry : The entry point is CHCIM1.
Routines Called :
• Real and Integer Input Conversion
(CHCII)
• PRMPT (CZATJ1)
Error Checks : If the complex data field in
the input buffer contains no real field or
only one real field, or if there is a mis-
CHCIN — Complex Output Conversion
This routine converts a complex list
item consisting of two single- or double-
precision floating-point items to a complex
data field in an output buffer. Each
floating-point list item is converted to a
real data field according to the format
code Fw.d, Ew.d, Dw.d, or Gw.s, where w is
the real field width, d is the width of the
decimal fraction, and s is the number of
significant digits. (See Chart IN.)
Entry : The entry point is CHCIN1.
Routine Called : General Output Conversion
(CHCIT)
Error Check : If the FORMAT specification
(F, E, D, or G) is improperly given or the
real data field is greater than the permis-
sible range, CHCIT issues an error message.
Operation : CHCIN passes the first list
item as a parameter to CHCIT for conversion
to a real data field. When CHCIT completes
the conversion process and returns to
CHCIN, CHCIN passes the second list item to
CHCIT for conversion, completing the com-
plex output conversion.
CHCIO — Alphameric and Hexadecimal Input
Conversion
This routine transfers a specified num-
ber of bytes (alphameric or hexadecimal
characters) from an input buffer area to a
list item. The format is Aw (alphameric)
or Zw (hexadecimal) , where w, field width,
is the number of characters being trans-
ferred. (See Chart IO.)
Section 5: I/O Routine Descriptions 39
Entry ; The entry points are CHCIOl
(alphameric data) and CHCI02 (hexadecimal
data) .
Operation : When the conversion format is
alphameric, this occurs:
* If the byte size of the list item
equals the number of alphameric charac-
ters in the buffer, a simple move is
executed, with no data test made*
* If the byte size of the list item is
less than the number of alphameric
characters in the buffer, the excess,
left-most characters are truncated in
the list item field,
* If the byte size of the list item is
larger than the number of alphameric
characters in the buffer, the charac-
ters are transferred into the list item
field, left™ justified.
When the conversion format is hexadeci-
mal, the field is scanned for valid hexa-
decimal characters. If invalid, a call is
made to PRMPT. If valid, the characters
are converted to binary and moved to the
input item. If the input field was pre-
ceded by a minus sign, the result is com-
plemented. In either case, a return is
made to the calling routine,
CHCIP — Alphameric and Hexadecimal Output
Conversion
This routine transfers a list item con-
taining a specified number of bytes
(alphameric or hexadecimal characters) to
an output buffer area. The format is Aw
(alphameric) or Zw (hexadecimal) , where w,
field width, is the number of characters
being transferred. (See Chart IP.)
Entry : The entry points are CHCIP1
(alphameric data) and CHCIP2 (hexadecimal
data) •
When the list item contains hexadecimal
data, it is unpacked and translated to hex-
adecimal digits in the output buffer.
C HCIQ — Logical Input (.:■ •aversion
This routine converts a logical field in
the input buffer area. The field has the
format Lw, where w is the field width.
(See Chart IQ. )
Entry : The entry point is CHCIQ1.
Operation : The list item is set to an
internal value of .FALSE. (binary zero).
The logical data field is then scanned
until either an F or a T is encountered.
In the first case, immediate return occurs.
In the second case, the list item is set to
an internal value of .TRUE. (binary one)
before returning.
CHCIR — Logical Output Conversion
This routine converts a list item to a
logical field in the output buffer area.
The field has the format Lw, where w is the
field width. (See Chart IR.)
Entry : The entry point is CHCIR1.
Operation : The output buffer area is first
filled with blanks. If the list item con-
tains a value of zero, the letter F is
placed in the last buffer area position;
otherwise, the letter T is inserted.
CHCIS — General Input C aversion
This routine converts a data field in
the input buffer to a li.t item according
to the format Gw.s, where w is the field
width and s is an optional specification of
the number of significant digits. (See
Chart IS.)
Operation : When the list item contains
alphameric data, this occurs:
• If the byte size of the list item
equals the size of the buffer area, a
simple move is executed, with no data
test made.
» If the byte size of the list item is
less than the size of the buffer area,
the list item is entered right-
justified and padded to the left with
blanks .
• If the byte size of the list item is
greater than the size of the buffer,
the list item is transferred to the
buffer area left- justified with excess
characters truncated.
Entry : The entry point is CHCIS1.
Routines Called :
• Real and Integer Input Conversion
(CHCII)
• Logical Input Conversion (CHCIQ)
• Alphameric and Hexadecimal Input Con-
version (CHCIO)
Error Checks : CHCIS does no error check-
ing. Error checks, if any, are made by the
called data conversion routines.
O peration : If the conversion type is *N' f
indicating a call from the NAMELIST Proces-
sor (CHCID) , this happens;
HO
Type of Data
J
Integer
CHCII
Real
CHCII
Logical
CHCIQ
Alphameric
CHCIO
Hexadecimal
CHCIO
• If the data field begins with a quote,
a literal character string is moved to
the list item,
• If the data field begins with a • Z' ,
hexadecimal conversion is performed.
If the conversion type is not f N' , the
list item is checked to determine which
conversion routine should be called:
Routine Called
(Entry point CHCIG)
(Entry point CHCIK)
(Entry point CHCIQ1)
(Entry point CHCI01)
(Entry point CHCI02)
CHCIT — General Output Conversion
This routine converts a list item to a
data field in the output buffer, according
to the format Gw.s; where w is the field
width and s is an optional specification of
the number of significant digits. (See
Chart IT.)
Entry ; The entry point is CHCIT1.
Routines Called ;
• Integer Output Conversion (CHCIH)
• Real Output Conversion (CHCIJ)
• Logical Output Conversion (CHCIR)
Error Checks ; CHCIT performs no error
checks. Discrepancies between the size and
type specification of the list item and the
data field are detected by the called con-
version routine.
Operation ; CHCIT examines the list item
type. If logical, a call is made to CHCIR.
If integer, a call is made to CHCIH. If
real, the magnitude of the item is tested.
Should it fall in the range 0.1 to 10**s, a
call is made to CHCIJ using F format speci-
fication. If real, and outside this range,
a call is made to CHCIJ using E or D format
specification.
CHCIU — List Termination
This routine terminates list processing
for a READ, WRITE, PRINT, or PUNCH state-
ment, and completes any I/O operation that
is pending. (See Chart IU.)
Entry : The entry point is CHCIU1.
Routines Called:
• Format Processor (CHCIF)
• I/O Control (CHCIO
The final return is made with registers
unchanged, except that register 13 is set
to the address of the calling module's
PSECT and register 15 is set to zero.
Operation : If the previous list processing
was an unformatted READ, no further action
is necessary, and control returns to the
caller. If the previous action was a for-
matted READ, CHCIF is entered with a zero
list item to indicate termination of the
FORMAT scan. This permits scanning for
Hollerith format controls that have still
not been processed, or skipping records
until the end of the format character
string.
If the previous action was an unfor-
matted WRITE, a final call is made to CHCIC
to write the complete logical record.
If, the previous action was a formatted
WRITE, CHCIF is entered as above, followed
by CHCIC.
CHCIW
Exit
CHCIW is described in "Section 4: Ser-
vice Subprogram Descriptions."
CHCBD — I/O Interruption and Machine
Indicator Routine
At the beginning of all FORTRAN main
programs, the compiler generates code that
calls this routine at entry point CHCBD1 .
Entry : The entry point is CHCBD1.
Operation : This routine sets bits in the
PSW such that the fixed-point overflow and
significance exceptions will be ignored,
and directs the system interruption handler
as to where to pass control if any of the
following exceptions occur:
Exception
Specification
Exponent Overflow
Exponent Underflow
Divide Check
Subprogram
CHCBE2
CHCBE3
CHCBE4
CHCBE5
In addition, this routine initializes the
machine indicator flags and the sense light
simulators, and clears any pointers to
entries in the DCB table. It then returns
control to the calling program.
Section 5: I/O Routine Descriptions 41
SECTION 6: FLOWCHARTS
The flowcharts in this manual have been produced by an IBM program, using ANSI sym-
bols. The symbols are defined in the left column below, and examples of their use are
shown at the right.
DEFINITION
TERMINAL B
LOCK J
PROCESS BLOCK
INDICATES AN ENTRY OR
TERMINAL. POINT IN A FLOW-
CHART; SHOWS START, STOP,
HALT, DELAY, OP INTERRUP-
TI >N . MAY' AIS<, ' NL>I< AT!
RETURN TO THE CALLING
INDICATES A PROCESSING
FUNCTION OR A DEFINED OP-
ERATION CAUSING CHANGE IN
VALUE, FORM OR LOCATION
OF INFORMATION.
DT \
INDD'ATF'-' A
DM' IS ION OP
"~<a r"> H \
DECISION ^V
THAT DI-OI-OM
■ > ', a'h :, 'i
BLOCK ~y
> A NUMBFT -if
A. " 1 t\AT:
PATHS SP< l.L
- M Im;i *l
INDICATES A SUBROUTINE i
MODULE THAT IS DESCRIBE!"
IN THIS manua;
I . ; -VNAME J
MODNAME IS THE LcjAD
NAME OF THE ROUTINE
FLOWCHART.
uTHERMuD
'" 'NTRuL 'P
''HARTS .
OF THE FIRST
oGRAM EXECUTION
-OU O L-l : r. • ■ !
OCK E3 WHEN THE
LABEL2 IS THE LABEL • ■¥
CODE IN THIS ROUTINE- I-
IS PASSED To THE SUBPo
RETURNS TO THE NEXT IN
INC THE SUBROUTINE OAI
EN'IRYPT IS THE ENTRY P
VIA: PAHSMEOH
/INPUT OUTPUT
PREPARATION
ARATION \
MODULE
IN THE
OTHER M
INDICATES A
PROCESS THAT
CHANGES SYS*
LM OPERATION
FOB EXAMPLE
SETS A SWI1
MODIFIES AN
incfo regis:
OR INITIAI t.
FS A ROUT INF
INDICATES ENTRY TO C
FROM ANOTHER BLOCK C
SAME FLOWCHART PAGE .
INDICATES ENTRY TO OP EX!
FROM A BLOCK ON ANOTHER
PAGE OF THE SAKE SET t >F
FLOWCHARTS .
PP=ENTPYPT
CHART AC
VIA: PASSMECH
LABEL3 IS THE LABEL T
CODE IROM WHICH *<•>,". b>
'PREDEFINED PROCESS PLY
DOCUMENTED IN ANOTHER
(-PDPNM- MAY ALSO BE U
ING BLOCK) .
i xt< u'l ion < ntimje' - .-. . : , J
THE DECISION IS YES. up
PAGE 2 OF THIS SET 6F FI
THE DECISION IS NO.
THE t 'FFPAOl U^NNF.-'I.-R *'/■
»'ATES THAT EXECUTE N N
H3 FROM ANOTHER PAGE ' T
'"PARTS . 1 -ilS C'CCr
WITH THE ON PAGE CONNF- |-.
42
Program Logic Manual
GY28-2020-2
FORTRAN IV Library Subprograms
Flowcharts on pages 43-88 were not scanned.
APPENDIX A; FORTRAN DATA MANAGEMENT
This appendix describes the assumptions
that the FORTRAN I/O library makes in
initializing DCBs with information concern-
ing record format (RECFM) and data set
organization (DSORG) . These asumptions are
described in this appendix to help reduce a
frequent source of error encountered by the
user when performing I/O.
Introductory material is presented on
the DCB describing its general use, con-
tents, and sources of initialization,
before discussing the permissible record
formats and data set organizations, (For a
description of the DCB in terms of its
size, format, and use by the rest of TSS/
360, see System Control Blocks PLM . )
DCB Use
The Data Control Block CDCB) is created
by DCB Management (CHCIB) and is used by
certain data management routines which are
invoked by macro instruction references in
I/O Control (CHCIC) . The DCB is required
for all I/O performed using either BSAM or
VAM. However, the DCB is not required for
I/O when using the GATE macro instructions,
even though CECIB reserves space for one.
In this case, the DD^AME field is used to
save the name that the data set would have
had if GATE were not used„ The principal
reason for constructing this dummy DCB for
GATE is to retain consistency for various
routines when handling data set reference
numbers with no corresponding DDEF
commands •
DCB Content
The DCB contains information such as the
DDNAME, type of data set organization, the
type and size of records, block size for
blocked data sets, number of buffer areas,
exits for SYNAD and EODAD, and various con-
trol flags used by data management,
DCB Initialization
The FORTRAN I/O routines, when proces-
sing an input data set, take advantage of
information in the DCB to adapt to the
characteristics of the data set and read it
correctly* Characteristics are based on
the parameters for a DCB that can be sup-
plied from:
• The user program — type of I/O used
and associated data format,
• User-supplied DDEF commands — some of
the information in the DCB can be
changed in this manner; however, the
extent of change is limited.
• Input data set labels — these override
both of tha above sources of informa-
tion, within limits set by data
management.
Combinations of DSORG and RECFM
Table 12 gives the permissible combina-
tions of record formats and data set
organizations that may be specified when
using the FORTRAN I/O library.
Table 12.
Combinations of DSORG and RECFM
values
A
N
i n
I A
I N
I N
I N
| N
| N
| N
| N
I L
Codes mean:
A - Acceptable
L - Limited acceptable
N - Not acceptable
DSORG abbreviations mean:
VS -
RECFM (- r
V
VB
VT
F
FB
FS
FT
FBS
FBT
FBST
FST
U
| VS
-+
1
DSORG VALUES
PS I VSP 1 VI | VIP
JL-
A
N
N
A
N
N
N
N
N
N
N
N
1 A 1
A I
A |
1 A 1
N
N |
1 A
| N
N |
1 A 1
A
A |
1 A
N
N |
t A
N
N [
1 A
| N
1 N |
I A
N
N I
1 A
| N
N |
1 A
| N
N |
1 A
1 N
1 N 1
1 A
i L
| N |
H
Virtual sequential (direct access
only)
Physical sequential (any device
except terminals)
Virtual sequential partitioned
(direct access only)
Virtual index sequential
(direct access only)
Virtual index sequential partition-
ed (direct access only)
RECFM abbreviations mean:
Variable-length unblocked records
Variable-length blocked records
Variable- length unblocked with
track overflow
Fixed- length unblocked records
Fixed-length blocked records
Same as F, no truncated blocks or
unfilled tracks
Same as F, track overflow
Same as FB, no truncated blocks
or unfilled tracks
Same as FB, track overflow
FBST- Same as FBS, track overflow
FST - Same as F, no truncated blocks,
track overflow
Undefined record length
PS -
VSP -
VI -
VIP -
V
VB
VT
F
FB
FS
FT
FBS
FBT
U
I
Appendix A: FORTRAN Data Management 89
Any of the RECFM codes shown can be fol-
lowed by the letter A or the letter M. A
indicates that the first character of every
logical record is a FORTRAN carriage or
punch control character, M indicates that
the first character of every record is a
TSS/360 machine control byte. In general,
the M option cannot be used by FORTRAN out-
put data sets, since the control codes are
unprintable and do not conform to FORTRAN
conventions*
records may span over data management phys-
ical records. In the management of unfor-
matted FORTRAN data, the first two bits of
every VS physical record or the third byte
of every PS physical record is a control
byte defined as follows;
X^O 1 A FORTRAN logical record does
not span into or out of the
data management physical
record.
Unformatted FORTRAN Logical Records
Under any of the organization techniques
used, an unformatted WRITE statement may
lead to a logical record that exceeds the
length of the maximum record supported by
the access method. Furthermore, it is not
possible to enter the byte size of the
entire FORTRAN logical record into the
beginning of the I/O physical record
without the possibility of tying up an
indefinite amount of virtual storage.
Therefore, unformatted FORTRAN logical
X'Ol* This data management physical
record is the first of a span.
X'0 2 1 This data management physical
record is the last of a span.
X*03 f This data management physical
record is within the range of a
span.
No data management logical record will
be written containing more than one unfor-
matted FORTRAN logical record.
90
APPENDIX B: EXTERNAL NAMES SUMMARY
All FORTRAN library routines have five-letter names beginning with the letters 'CHC.
The names of mathematical library routines begin with ' CHCA 1 or , CHCB I f and the names of
I/O library routines begin with • CHCI' except one — CHCBD. All other external names
consist of the routine names with suffixes.
Table 13. External names of FORTRAN IV library subprograms
j Mathematical
I Subprograms j
Service Subprograms
h
Entry | See Tables 2 9 3 f |See Tables 7 and 9, and
Name (and 7. (Section 4,
I/O Subprograms
See Section 5.
Routine [See Tables 2 f 3, [See Tables 7 and 9, and
Name (and 7. (Section **.
CSECT (Routine name (CHCBE: Routine name suffixed
Name (suffixed by ^".(by 'W. CHCIV and CHCIW:
j (Routine name suffixed by f C f .
See Section 5,
Routine name suffixed by ' C.
H
H-
-H
PSECT (Routine name (CHCBE: Routine name suffixed
Name (suffixed by ' R'.(by , R f . CHCIV and CHCIW;
| (Routine name suffixed by 'W.
A-_. . X . .
Routine name suffixed by ' W
Appendix B: External Names Summary 91
INDEX
Where more than one page reference is
given* the major reference is first*
.FALSE logical constant 41
.TRUE logical constant 41
A conversion code 40
Absolute value subprograms 19,4
Access methods
basic sequential CBSAM) 34 r 35
virtual CVAM) 34,35
Access qualifier 34
Accuracy figures , mathematical
subprograms 15-24
ALGAMA subprogram 19,5
Algorithms of mathematical subprograms 14
ALOG subprogram 15,4
ALOG10 subprogram 15,4
Alphameric and Hexadecimal Input Conversion
CCHCIO) 40,41
calls from other routines
CHCIF 37,38
CECIS 41
Csee also I/O library, overview)
flowchart 80
Alphameric and Hexadecimal Output
Conversion CCHCIP) 41
calls from other routines
CHCIF 37, 38
CHCIS 41
Csee also I/O library, overview)
Arccosine and arcsine subprograms 15,4
ARCOS subprogram 15,4
Arcsine and arccosine subprograms 15,4
Arctangent subprograms 15,4
Argument ranges, mathematical
subprograms 15-23
Arguments, mathematical subprograms 14-23
Arrays
dimensions 36
elements 35-37
length 36
names 35, 36
size 36
subdivision, levels of 3 6
subscripting 36
ARSIN subprogram 15,4
Asterisks 40,41
ATAN subprogram 15,4
ATAN2 subprogram 15,4
Attributes
I/O routines 30
mathematical subprograms 14
service subprograms 26
BACKSPACE operation 13,1
CHCIA 31,32
CHCIC 33-35
Csee also I/O library, overview)
Basic Sequential Access Method
CBSAM) 32,34,35
Binary list item 39
Blanks 39
Blocked records 34,35
BSAM Csee Basic Sequential Access Method)
BSP macro instruction 3 5
Buffering 34-42
CABS subprogram 19, 4
CALL statement 1
Calling relationships
I/O library 3-13
mathematical library 4 , 5
Carriage control 90
CCOS subprogram 17,5
CDABS subprogram 19,4
CDCOS subprogram 15,5
CDEXP subprogram 15,5
CDLOG subprogram 15,4
CDSIN subprogram 15,5
CDSQRT subprogram 15,4
CEKBF routine 38,39
CEKT1 ... CEKT6 macro instructions 25
CEKZA macro instruction 25
CEKZD macro instruction 25
CEXP subprogram 15,5
Character string 38-42
CHCAA routine 15, 4
CHCAB routine 15,4
CHCAC routine 15,5
CHCAD routine 15,5
CHCAE routine 15,4
CHCAF routine 15,4
CHCAI routine 17,5
CHCAJ routine 17,5
CHCAK routine 19,5
CHCAL routine 19,5
CHCAM routine 15,5
CHCAN routine 15,5
CHCAO routine 15,4
CHCAP routine 15,4
CHCAQ routine 15,17,5
CHCAR routine 15,5
CHCAS routine 15,4
CHCAT routine 15,4
CHCAU routine 19,4
CHCAV routine 19,4
CHCAW routine 15,4
CHCAX routine 15, 4
CHCAY routine 17,4
CHCAZ routine 17,4
CHCBA routine 17,5
CHCBB routine 17,5
CHCBC routine 21,25
mathematical library overview 5
CHCBD Csee I/O Interruption and Machine
Indicator Routine)
CHCBE Csee Interrupt and Machine
Indicator Routine)
CHCBG routine 21,25,5
92
CHCBH routine 21,25,5
CHCBI routine 21,25,5
CHCBJ routine 21,25,5
CHCBK routine 21,25,5
CHCBM routine 21,25,5
CHCBQ routine 15, 4
CHCBR routine 15, 4
CHCBT routine 19,5
CHCBU routine 19,5
CHCBV routine 19,5
CHCBW routine 19,5
CHCBZ (see Error Processor)
CHCIA (see I/O Initialization)
CHCIB (see DCB Maintenance)
CHCIC (see I/O Control
CHCID (see NAMELIST Processor)
CHCIE (see List Item Processor)
CHCIF (see FORMAT Processor)
CHCIG entry point 42
CHCIH (see Integer Output Conversion)
CHCII (see Real and Integer Input
Conversion)
CHCIJ (see Real Output Conversion)
CHCIK entry point 42
CHCIL1 entry point 40
CHCIM (see Complex Input Conversion)
CHCIN (see Complex Output Conversion)
CHCIO (see Alphameric and Hexadecimal
Input Conversion)
CHCIP (see Alphameric and Hexadecimal
Output Conversion)
CHCIQ (see Logical Input Conversion)
CHCIR (see Logical Output Conversion)
CHCIS (see General Input Conversion)
CHCIT (see General Output Conversion)
CHCIU (see List Termination)
CHCIV (see Dump Routine)
CHCIW (see Exit Routine)
Class (see also Type)
of variable 36
CLOG subprogram 15,4
CLOSE macro instruction 35
Closing parenthesis 39
Combinations of DSORG and RECFM 89
Common and natural logarithm
s ubpr o grams 15,4
Compiler 31,35
Compiler- generated entry names 25,2
Complemented error function
subprograms 19 , 5
Complex Input Conversion (CHCIM) 40
calls from other routines
CHCID 35
CHCIF 37,38
(see also I/O library, overview)
flowchart 78
Complex list item 40
Complex Output Conversion (CHCIN) 4
calls from other routines
CHCID 35
CHCIF 37,38
(see also I/O library, overview)
flowchart 76
Complex result, how passed 14
Computations of mathematical
subprograms 15-23
Control Initialization
(see I/O Initialization)
Conversion codes 39,3
data conversion routines, use by 39-4 2
Conversion routines
(see Data conversion routines)
COS subprogram 17,5
COSH subprogram 17,5
COTAN subprogram 17,4
CSECTs 89,23,25
CSIN subprogram 15,5
CSQRT subprogram 15,4
CZAEB routine 32,33
CZAFQ1 routines 30
CZCGA routine 32
d Field 38-40
D conversion code 39,40
DARCOS subprogram 15,4
DARSIN subprogram 15, 4
Data Control Block (DCB) 32,33,3
CHCIA 31
CHCIC 33-35
CHCIW 29,30
contents 89
initialization 89
prefix 32,34
table used by CHCIC 35,43
use 89
Data conversion routines 35,37
parameter lists 30,31
(see also I/O library, overview; CHCIH;
CHCII; CHCIJ; CHCIM; CHCIN; CHCIO;
CHCIP; CHCIQ; CHCIR; CHCIS; CHCIT)
Data event control block (DECB) 32
Data field, I/O 38-42
Data management routines 32,33,3,6
Data set
access qualifier 34
label 89
read-only 34
reference number 31-35
DATAN subprogram 15,4
DATAN2 subprogram 15,4
DCB (see Data Control Block; DCB
Maintenance)
DCB Maintenance (CHCIB) 32-34
calls from other routines
CHCIA 31
CHCIC 33, 34
CHCIW 30
(see also I/O library, overview)
flowchart 49
DCOS subprogram 17, 5
DCOSH subprogram 17,5
DCOTAN subprogram 17,4
DDEF command 32-34,3
DDNAME 89
DECB (Data Event Control Block) 32
Decimal conversion 39
Definitions of subprogram
computations 15-23
DERF subprogram 19,5
DERFC subprogram 19,5
DEXP subprogram 15,4
DGAMMA subprogram 19, 5
Digit 39-42
Dimension 36
Dimension product 36
Index 93
Direct -reference mathematical
subprograms 11-25,1,2
Direct-reference service subprograms
11-25,1,2
Divide Check Handier 26-28
Divide Check Tester 26-28
DLGAMMA subprogram 19,5
DLOG subprogram 15 , 1
DLOG10 subprogram 15,4
Double buffering 35
Doubl e- preci s ion
argument (notation) 2 3
complex argument (notation) 23
list item 39,10
DSIN subprogram 17 f 5
DSINH subprogram 17,5
DSORG values 32,89
DSQRT subprogram 15, 1
DTAN subprogram 17,4
DTANH subprogram 19,5
Dummy DCB 33
Dump Routine (CHCIV) 28-29,26
flowchart 87
(see also I/O library, overview)
DUMP subprogram 28,29
(see also Dump Routine)
CHCIA 31
CHCIB 3 2
CHCIC 33,34
CHCID 35
CHCIE 37
CHCIF 37
FORTRAN compiler (user program) 3
(see also I/O library, overview)
1. lowchait 8
EXIT subprogram 29 f 3Q
(see also I/O library, overview)
EXP subprogram 15,4
Exponent, I/O involving 39, 40
Exponent Overflow Handler 26-28
Exponent Overflow and Underflow
Tester 26-28
Exponent Underflow Handler CCHCBD1) 26-28.
Exponentiation subprograms 21
ent ry par amet er s 11
(see also CDEXP subprogram? CEXP
subprogram; DEXP subprogram; EXP
subprogram; Indirect-reference
mathematical subprograms; Mathematical
1 i b r u r y , o v e r v i e w )
External entry name (definition) 2
External names 91
E conversion code 39-41
Elements, array 35,37
END exit 30-33
control flag in DCB prefix 3 2
End file lockout indicator 31,35
END FILE operation l r 13
CHCIA 31 r 32
CHCIC 33-35
Csee also I/O library, overview)
Entry names
I/O subprograms 91
mathematical subprograms 15-25
service subprograms 26-30 , 2
EODAD exit 33 f 89
Equal sign 35
Equations showing subprogram
computations 15-23
ERF subprogram 19 , 5
ERFC subprogram 19 , 5
ERR exit 30-33
control flag in DCB prefix 32
Error checks
mathematical subprograms 11-23
service subprograms 26-29
Error conditions (see Error checks)
Error function subprograms 19 , 5
Error Processor CCIiCBZ, math
library) 25 r 23
calls from other routines 11,2
CHCBE 28
Csee also Mathematical library,
overview)
flowchart 47
Error routines
CHCBZ 25,23
Exceptions 26,28
Exception Processing Enabler
CCHCBD1) 26-28
Exit Routine CCHCIW) 29-35,26,27
calls from other routines 30
F conversion code 39-11
FCDXI entry name 2 5
FCDXJ entry name 2 5
FCXPI entry name 2 5
FCXPJ entry name l",
FDXPD entry name 2 5
FDXPI entry name 25
FDXPJ entry name 25
Field width, I/O 38-42
Fill characters 34
FIXPD entry name 25
FIXPI entry name 25
FIXPJ entry name 2 5
FIXPR entry name 25
FJXPD entry name 2 5
FJ X P I e n t r y na me 2 5
FJXPJ entry name 25
FJXPR entry name 2 5
Floating-- point, list item 39,10
Flowcharts 12-88
Format
controls, Hollerith 12
of FORTRAN logical records 8 8
of NAM EL 1ST output 36
FORMAT
(set also FORMAT Processor)
clo.-;in<-; paieiiUiCwiS 3 C J
cedes 38-12
control 37-12,1
37,38
31,32
32
36, 37
37-39
10 I/O library, overview; Data
CEKBF
CHCIA
CHCIB
CHCIE
CHCIF
(see a Is
Conversion routines)
conversion code 39-12
parenthesis 38 r 39
repetition factor 38
scale factor 3 9
91
statement, operations involving
(see FORMAT control)
string 37,38
table 38 f 39
FORMAT Processor (CHCIF) 37-39
calls from other routines
CHCIA 31 f 32
CHCIE 36 f 37
CHCIU 42
(see also I/O library, overview)
flowchart 69
Format-V records 34
Formatted READ 37-42,8
Formatted WRITE 37-42
with list 10
without list 11
FORTRAN
compiler 31, 35
CEKBF routine 37,38
SYSPFMT entry point to CEKBF 37,38
data management 87,88
combinations of DSORG and
RECFM 87,88
logical records 88
unformatted logical records 88
(see also Data Control Block)
source statements 14
Fraction- I/O involving 39,42
FRXPD entry name 25
FRXPI entry name 25
FRXPJ entry name 25
FRXPR entry name 25
Functional flow
I/O library routines 7
mathematical library subprograms 4,5
Function values of mathematical
subprograms 14-23
Functions
mathematical subprograms 14-21
service subprograms 26-29
G conversion code 40-42
GAMMA subprogram 19,5
GATE
DCB 33
I/O 29,33,34
macro instruction facility 29,33,34
GATRD macro instruction 3 4
GATWR macro instruction 33,34
General Input Conversion (CHCIS) 41
calls from other routines
CHCID 35
CHCIF 37,38
(see also I/O library, overview)
flowchart 81
General Output Conversion (CHCIT) 4 2
calls from other routines
CHCID 35,36
CHCIF 37,38
CHCIN 40
(see also I/O library, overview)
flowchart 82
GET macro instruction 34
GTWRC macro instruction 34
H conversion code 38
Header and trailer records in NAMELIST
output 36
Hexadecimal input conversion (see
Alphameric and Hexadecimal Input
Conversion)
Hexadecimal output conversion (see
Alphameric and Hexadecimal Output
Conversion)
Hollerith format controls 42
Housekeeping functions, I/O 43
Hyperbolic sine and cosine
subprograms 17,5
Hyperbolic tangent subprograms 19,5
I conversion code 38, 39
I/O Control (CHCIC) 33-35
calls by other routines 30
CHCIA 31,32
CHCID 35
CHCIE 37
CHCIF 37
CHCIU 42
CECIW 30
(see also I/O library, overview)
flowcharts 50-67
I/O Initialization (CHCIA) 30-32
called by FORTRAN compiler 35,66
flowchart 50
(see also I/O library, overview)
I/O Interruption and Machine Indicator
Routine (CHCBD) 41
I/O library 30-42,2-13
calling relationships 6,7
housekeeping functions 42
internal relationships 3-13
overview 2,3-13
subdivisions 3
(see also I/O routines)
I/O list control routines 3,7
I/O operation control routines 3,7
I/O operations (see BACKSPACE operation;
END FILE operation; PRINT operation? PUNCH
operation; READ operation; REWIND
operation; WRITE operation; WRITE
Initialize operation)
I/O PSECT 42
I/O routines 30-42,2-13
attributes 30
names 30,89
P SECTS 30
save areas 30
work areas 30
(see also Data Conversion routines)
I/O services control routines 3,7
I/O statement (see I/O operations)
Indirect-reference mathematical
subprograms 25 ,1 , 2,14
Indirect-reference service subprograms 2
INOUT (OPEN option) 34
INPUT (OPEN option) 34
Integer input conversion (see Real and
Integer Input Conversion)
Integer Output Conversion (CHCIH) 39
calls from other routines
CHCIF 37,38
CHCIT 42
(see also I/O library, overview)
flowchart 71
Index 95
Integer result, how passed 14
Interrupt and Machine Indicator Routine
CCHCBE) 26-28,4
flowcharts 44-46
Item length 37
Item type 15-23,36-41
JFCB (Job File Control Block) 33
KEYLEN values 32
L conversion code '41
Label, data set 89
Left parenthesis 35
Linkage
I/O routines 30
(see also I/O Communication)
mathematical subprograms 14
service subprograms 26
(see also Calling relationships)
List control 2,3
flag in DCB prefix 32
(see also I/O library, overview; List
Item Processor)
List item 36-42
binary 39
complex 40
double- precis ion 39 f 4
floating-point 39,40
single-precision 39
List Item Processor CCHCIE) 30,36-37
called by FORTRAN compiler 31
Csee also I/O library, overview)
flowchart 70
List processing Csee List control)
List Termination CCHCIU) 30,42
called by FORTRAN compiler 31
flowchart 86
Csee also I/O library* overview)
Literal" character string 42
Log-gamma and gamma subprograms 19,5
Logical constats 41
Logical Input Conversion (CHCIG) 41
calls from other routines
CHCIF 37,38
CHCIS 41,42
(see also I/O library, overview)
flowchart 82
Logical Output Conversion CCHCIR) 41
calls from other routines
CHCIF 37,38
CHCIT 42
(see also I/O library, overview)
flowchart 83
Logical record 90
CHCIC 34
CHCID 35
CHCIE 37
CHCIF 39
Macro instructions
BSP 33,35
CEKT1. . .CEKT6 25
CEKZA 25
CEKZD 25
CLOSE 35
GATK facility
(JATWR 33,34
GATRD 33,34
GTWRC 3 4
GET 33,34
in I/O library 3 3
in mathematical library 25
NOTE 33,35
OPEN 34,35
POINT 33,35
PUT 33,34
READ 33
SAVE 25
SETL 33-35
STOW 33
WRITE 33,34
Mathematical library 14-25,2-5
ca 1 1 i ng rela t i uns h I. us 4 , 5
i n 1 1 .• i na 1 re 1 at i on sh i p s 2-5
overview 2-5
s u b divisions 2
Ma t heiua tica 1 sub pr 09 ra ma 14-2 5 , 1
a ecu racy f igur es 1 5-19, 23, 2 4
arguments 15-23
attributes 14
computations l\>--23
definitions ( descript ive
eq ua t i on s ) 1 L ) - 2 3
d i r e ct -referenc <= 15-2 b
e n t r y name s 15-25,91
e nt 1 y paramete r s 1 4
equations 15-23
error action 14
exit parameters 14
functions 14-21
i nd i r ect-refere nee 2 3-25
int i'-r relationships 4 , 5
overview 3-5
pui p-^ses 15-23
storage estimates 15-23
tables 4,5,15-21
Minus sign, I/O involving 39- m
n field' 39
N conversion code u2
NAM EL I ST
control* 1,3
CHCIA 31,32
CIIC IB 32
CHCIC 3 4
CHCID 35 # 36
CilCIS- 32
(see also I/O library, overview)
d im ension prod u c t 3 6
format for our put 3 6
•header and trailer records 36
input 36
name 35, 36
notation 36
out put 3 6
record 35
s tat ement ( s et< NAMKL1S T con tro 1 ;
t a b i e 3 5 f 3 6
• 1 1 a tier and heu d ex records 3 1
96
variable 36
(see also NAMELIST Processor)
NAMELIST Processor (CHCID) 35,36
called by CHCIA 31,32
Csee also I/O library, overview)
flowcharts 68-69
Names
I/O routines 30 # 91
mathematical subprograms 14-23,91
service subprograms 27,91
Natural and common logarithm
subprograms 15,4
Non-format -V records 34 i
Notation, NAMELIST 36
NOTE macro instruction 35
OPEN macro instruction
Output buffer 34
Overview 2-13
34,35
Packed decimal conversion 39
Padding 34,39
Parameter lists (general information)
I/O routines 30
mathematical subprograms 14
service subprograms 26,28-30
Parenthesis 35,38,39
PAUSE subprogram 29-30
(see also Exit Routine; I/O library,
overview)
PDUMP subprogram 29
(see also Dump Routine; I/O library,
overview)
Physical record 37,90
POINT macro instruction 35
Powers of ten, table 39
PRINT operation 7
CHCIA 31
CHCIE 36
CHCIU 42
(see also I/O library, overview)
Program interruptions 26
PSECTs
I/O routines 30,91
(see also I/O Communications)
mathematical routines 23,91
label generation 1
locations passed in register 13 25
PUNCH operation 7
CHCIA 31
CHCIE 36
CHCIU 42
(see also I/O library, overview)
PUT macro instruction 34
Quote 42
Range of data field 39, 40
READ macro instruction 33
Read-only data set 34
READ operation 31-42
formatted 8
with list 8,9
with NAMELIST 9
without list 8,10
unformatted 9,10
(see also I/O library, overview)
Real argument (notation) 23
Real and Integer Input Conversion
(CHCII) 39
calls from other routines
CHCIF 37,38 - ^
CHCIM 4
CHCIS 41, 42
(see also I/O library, overview)
flowchart 75
Real Output Conversion (CHCIJ) 40
calls from other routines
CHCIF 37,38 4
CHCIT 42 r . ;j
(see also I/O library ^ overview)
flowcharts 73,74
Real result, how passed 14
RECFM (see Record format)
Record format (RECFM) ^ r
DSORG considerations 8^^0,32
format- V 34 t \
non-format-V 34
Records
blocked 34,35
logical (see Logical record)
physical 90,37
unblocked 35
Register save areas (I/O routines) 30, %
Repeat constant 35
Repetition factor, FORMAT 38
Repetition notation for NAMELIST J/O 36
Return parameters (see Exit parameters)
REWIND operation 13 c ,
CHCIA 31,32 r
CHCIC 33-35
(see also I/O library, overview)
Right parenthesis 38,39
Routine names
I/O 91,30
(for list, see Table of Consents)
mathematical 91,14-21 > n£ &
for indirect-referent : , lco
subprograms 21,25 i-
service subprograms 26-29* t *
s field 40-42 u
Save areas for I/O routines 30
(see also I/O Coram#ni cation)
SAVE macro instruction 25
Scale factor, FORMAT 39
Sense light subprograms (SLITE,
SLITET) 26-28
Service subprograms 26-29,2 ^;
attributes 27
entry names 26,27
entry parameters 26
error checks 26-27
parameters 26
SETL macro instruction 35
Sign of number, I/O involving 39-41
Significant digits 42
Simple variable 36
SIN subprogram 17,5
Sine and cosine subprograms
hyperbolic 17,5
trigonometric 17,5,16
Index 97
Single-precision
a rgument ( not at i on > 2 3
list item 39
SINH subprogram 17 f 5
Size of data field 39,40
SLITE and SX»IT£T subprograms 26-28
Source program (see User program)
Spaces, in FORMAT I/O 3 8
(see also Blanks)
Spanning 34,90
Specification Exception Handler 26-28
Square root subprograms (CSQRT, CDSQRT,
DSQRT, SQRT) 15 f 4
STOP subprogram 29 f 30
(see also Exit Routine? I/O library,
overview)
Storage estimates
mathematical subprograms 15-23
service subprograms 27
STOW macro instruction 33
Subdivision, array 36
Subscripted variable 36
Subscripts 35,36
Supervisor 6
SVC instruction 30
SYNAD exit 33 f 89
Syntax of FORMAT string 37,38
SlfSIN 33
S3TSLIB 1
SYSOUT l t 3
CHCIB 33
CHCIW 29
CHCIX *2
SYSPFffT entry point to CEKBF 37 f 38
System entry na»e 21,23,25
Syste» interruption handler 2 8
T conversion code 38,41
Table of powers ten 40
TUN subprogram 17 f U
Tangent subprograms
hyperbolic 1 9 f 5
trigonometric 17, t*
TANH subprogram l i *,5
Task Supervisor 3*)
Ten, powers of* <-)
Seminal entry to HftMBLIST table 36
Trailer and header records in NAME LI ST
output 36
Trigonometric sine and cosine
subprograms 15-17,5
Trigonometric tangent subprograms 17,4
Truncation, in I/O 41
Type ( r ea 1 , intege r , e t c « ) 3 t - i l 2 t l *j -23
Unblocked records 35
Unformatted FORTRAN logical records 90
Unformatted READ 9,10,33
with list 9
without list 10
Unformatted WRITE 12
UPDATE (OPEN option) 3 4
User entry name 2 4
User program 30,36,37
diagram showing interface with I/O
library 6
VAM 34,35
Variable
class 15-23,36
location 36
name 35 f 36
NAMELIST 36
■simple 36
subscripted 36
type IS- 21, 36
Virtual Access Method (VAM) 34,35
VPAM data set 34
w field 39-43
width of data field 38-42
Work areas (see PSECTs)
WRITE Initialize operation 34,35
WRITE macro instruction 34
WRITE operation 31-42
•formatted 10, 11, 3 4
with list 10-12,36
with NAMELIST 11,34,36
without list 11,12
u n f o nna tt ed 12
Csee also I/O library, overview)
Z conversion code 41,42
Zoned decimal conversion 39
m
