BRITISH COMPUTER SOCIETY FORTRAN SPECIALIST GROUP
Minutes of a meeting held on
Friday, 9th November 1973 at
BCS Headquarters, 29 Portland
Place, London W1 at 10.30 a.m.
Present: Mr B.H. Shearing (Chairman) Alcock Shearing & Partners
Ms E. Aylmer-Kelly University of York
Mr P.D. Bond Philips Industries
Mr A. Bruce Government Communications
Mr F. Chambers John Harwell D.P. Ltd
Mr P.A. Clarke Rothamsted Experimental Station
Mr J.C. Cullen B.P.
Dr A.C. Day University College London
Mr D. Ellison University of Manchester R.C.C.
Mr J.C. Gilbert University of London C.C.
Mr P.J. Hammond B.P.
Mr J.P. Holland Tunnel Cement Ltd
Mr T.D. Palmer Compower Ltd
Mr K. St.Pier GEC Computers
Ms J.A. Smith A.W.R.E. Aldermaston
Mr P.J. Vyse Honeywell I.S. Ltd
Mr D. Winstanley University of Birmingham
Mr D.T. Muxworthy (Secretary) Edinburgh R.C.C.
Apologies
for absence Mr T.L. van Raalte A.W.R.E. Aldermaston
Action
1. APPROVAL OF The minutes of the meeting of the 27th July 1973
MINUTES were approved.
2. MATTERS ARISING The Chairman was waiting for the November edition BBS
FROM THE MINUTES of FORTREV, the working document of the American
Standards Fortran subcommittee X3J3, before
discussing its publication with the editor of the
Computer Journal.
3. BCS MISCELLANEA The attention of the meeting was drawn to the
following:
a. The Standards Committee of the BCS has forwarded a
copy of a Proposed Extended Mathematical Symbol
Definition for Interchange.
b. The IFIP congress in 1974 will be held in Stockholm
from August 5 to 10 and will include the MEDINFO
conference.
c. Contributions to 'Computing' from members of the
group on topics related to Fortran are welcomed.
Articles should be about 1000 to 1500 words long
and payment is made at the rate of £20 per thousand
words published.
d. Datafair produced a profit of £20,000 for the BCS.
e. The BCS Library is not to be run down.
f. A 'Computer Systems and Technology' Conference is to
be held in London on October 29 to November 1, 1974;
abstracts of contributed papers must be submitted by
January 7.
g. The list of BSI committees on which the BCS is
represented which was printed in 'Computing' of
August 9, 1973 was apparently not intended to be
exhaustive (the Programming Languages Committee
DPE/13 was omitted).
h. The IEE are to hold a debate on the motion, "That
this house believes that computer systems are not
organized for the benefit of the customer" during
the first half of 1974.
i. A new BCS specialist group is being set up; it is
called "Human aspects of computer management".
j. The BCS Objectives Review Committee held its first
meeting on November 6. Its terms of reference are
to define the short and longer term aims and
objectives of the BCS and to make recommendations
to Council at the April 1974 meeting. Written
contributions are welcome and may be addressed to
the chairman of the committee, G. Morris, or to the
Secretary of the Group, who is a member of the
committee.
k. Yet another error~ridden Fortran algorithm has been
published in the Journal (August 1973 pp 273-276)
and the Chairman and Secretary are to pursue the BHS/DTM
matter. Messrs Day and Winstanley offered to
referee the language aspects of algorithms written
in Fortran.
l. A BCS symposium to review command languages and
discuss JCL standardization will be held at the
City University on January 24, 1974.
4. IBM PROGRAMMING IBM presented their new programming language
LANGUAGE STRATEGY strategy at a recent user meeting attended by the
Secretary. The strategy is most favourable to
Fortran for it is no longer IBM policy to persuade
Fortran users to move to PL/I and IBM committed
themselves to support any new internationally
accepted standards in their six main languages
(APL, Basic, Cobol, Fortran, PL/I, RPG).
5. U.S. NAVY TEST The set of U.S. Navy programs, designed to test
PROGRAMS whether a compiler conforms to the ANSI standard
(v. minutes of 21.9.72, section 7) has been run by
Mr Clarke with the ICL System 4 compiler FTRAN3
(P8090) and with the Edinburgh R.C.C. compiler
FORTE. Errors were revealed in both compilers and
in the test programs themselves. Copies of the
summary report of these runs may be obtained from
Mr Clarke.
6. BSI STANDARDS Dr Day reported that the BSI DPE/13 Fortran sub-
ACTIVITIES committee had identified a number of typographical
errors in the ISO Fortran recommendation, ISO/R/1539;
the corrected document would be proposed as a
British Standard. (This is essentially the 1966
American Standard).
7. ECMA STANDARDS The ECMA Standards Committee has continued to meet
ACTIVITIES every two months to consider the X3J3 proposals.
They note that X3J3's work is scheduled until
September 1975.
8. MULTINATIONAL The attention of the meeting was drawn to the
DATA STANDARD Multinational Data Fortran Standard (Technical
Publication No. MD-STD 30-002 dated 23.10.72,
available from ICL) which describes, in the form
of the 1966 American Standard, the Fortran Language
intended to be implemented by ICL. It contains
some of the current X3J3 proposals. A large
number of errors had been found in the document by
Mr Clarke.
9. AMERICAN FORTRAN The minutes of the X3J3 meetings of July 9-12 and
STANDARDS September 11-14, 1973 were considered. The
ACTIVITIES principal decisions then taken were:
a. To clarify the input-output proposals.
b. To allow ENTRY and FORMAT to precede IMPLICIT.
c. To add colon, 'less than', 'greater than' to the
character set.
d. To use colon as the array bound separator in array
declarators, as the character position separator
in substring references and as the ellipsis
character in array block input-output.
e. To use two consecutive slashes as the concatenation
operator.
f. To allow character data in blank common provided
all blank common blocks in an executable program
contain only character data.
g. To adopt colon as the format scan terminator.
h. To allow array cross~sections to be used with sub-
string references.
i. To delete the requirement that each subprogram
contain a RETURN statement.
j. To require that an entry name be not used in an
executable statement before the ENTRY statement.
k. Not to allow < as an alternative to .LT. etc
(so that <, > are in the character set but have
no use).
l. Not to delete non-unity lower bounds for arrays.
m. Not to delete the SAVE statement.
Comments on these topics were sent to X3J3 and are
shown in appendix A to these minutes.
Mr Marwick had received a letter thanking him for
his free-format input-output report. A number of
flaws had recently been found in the X3J3 proposals
for list-directed input-output and the whole matter
is being rethought.
The schedule for accepting the new draft proposed
standard is:
a. X3J3 ballot. Minimum statutory time 30 days;
time proposed is 60 days. Most standards take two
or more such ballots.
b. X3 ballot to "accept for public review". Time
varies from 2 weeks to 30 days according to X3
meeting dates.
c. X3 Public Review. Four months are allowed.
d. X3 ballot. This takes 6 weeks from the end of
the public review, or from X3J3 completion and X3
approval of responses to all comments received
during comment period.
e. All negative X3 ballots are referred to X3J3 for
responses, which are again approved by X3. If the
responses involve technical change of substance,
there is an X3 30 day ballot to approve the
changes.
10. OTHER BUSINESS Members of the Group are invited to send to the
Secretary details of any papers or documents
related to Fortran they have produced, as
individuals or as Group members. This is to help
the BCS answer queries about specialized subjects.
11. DATE OF NEXT The next meeting will be held at 10.30 a.m. on
MEETING Friday, February 8, 1974 at BCS Headquarters.
DTM/PLU/1298
15th November 1973
To: X3J3
From: David Muxworthy, for BCS Fortran Specialist Group
Subject: Comments on X3J3/45 and X3J3/46.
1. (/45,4.1,FORTREV 13.13). Does an error condition occur if a format
field descriptor does not agree in type with the corresponding input-
output list item?
2. (/45,4.1). How is the record length specified when characters are
written to an unformatted direct-access file?
3. (/45,5.1). We are in favour of trailing comments.
4. (/45,4.2). We suggest that PARAMETER should be allowed before IMPLICIT,
in other words PARAMETER and IMPLICIT should have the same precedence.
5. (FORTREV 15.16g). ENTRY names within FUNCTION subprograms may be of
different type to the function. Is it allowed for one of these to be of
type CHARACTER and one to be another type?
6. (/46,4.11). We had a straw vote on adopting the alternate notation for
relational operators en bloc. It was rejected 5-9.