BRITISH COMPUTER SOCIETY - FORTRAN SPECIALIST GHDUP
MINUTES OF AGM HELD AT BCS HQ, MANSFIELD STREET, LONON ON 18 JUNE 1984
Present: Jackie Bettess - UC Swansea
Joyce Graves - University of Nottingham
David Muxworthy - University of Edinburgh
K Normington - Coventry Polytechnic
Mike Nunn - CCTA
T L van Raalte - MOD
John Reid - Harwell
Steve Watkins - UMIST
John Wilson - University of Leicester
Addresses: Chairman John Wilson
Computer Laboratory
University of Leicester
LEICESTER
LE1 7RH
Secretary Mike Nunn
CCTA
Riverwalk House
157 Millbank
LONDON
SW1P 4RT
Treasurer Keith Normington
Computer Centre
Coventry (Lanchester) Polytechnic
COVENTRY CV1 5PB
1. MINUTES OF PREVIOUS MEETING [16 April 1984]
Amendments should be made as follows:
Section 5(ii) first line
"David Muxworthy produced a summary report on the ISO Geneva meeting for the
BSI".
Section 6 page 3
" .... George Paul (IBM, X3J3 consultant)".
Section 6(v) page 5
"There was general support for BIT data type".
2. MATTERS ARISING
The draft BSI Standard document "Method of Specifying Requirements for FORTRAN
Language Processors" is presently being polished up by the BSI Editorial Department
and is expected to be released for public comment in mid-sumer. If all goes well
it could become a British standard in early 1985.
3. TREASURER'S STATEMENT
A copy of the Treasurer's accounts for the financial year ended 30 April 1984
is in appendix A. £550 was originally allocated to the Group by BCS to meet HQ
expenses but the cost of these services eventually came to £439.50. Main HQ
expense is sending out our newsletter. The Treasurer has written to ask them for
an analysis of their charges.
4. ELECTION OF OFFICERS
i. Officers were re-elected unanimously as follows:
Chairman - John Wilson
Secretary - Mike Nunn
Treasurer - T L van Raalte.(Due to personal circumstances Mr van Raalte has
resigned as Treasurer since the AGM. Keith Normington has
agreed to take over these duties).
Addresses of the above persons appear on the first page of these minutes.
ii. The Chairman felt there should be a vice-chairman in case he could not
attend a meeting at the last minute. This was agreed and Keith Normington
(Coventry Polytechnic) was elected vice-chairman.
5. CHAIRMAN'S BUSINESS
i. A report produced by the Chairman for BCS on Group activities over the
last year is in appendix B.
ii. The chairman proposed an increase in the Group's membership fee to £5 for
non-BCS members for the year ending April 1985. This was voted in favour
unanimously. The charge mainly covers the newsletter expense.
iii. Specialist Groups Board
- a new group called "Privacy and Security" has been formed
- "Distributed Database Working Group" is to become a group
- an "Electronic Publishing" group has just started up
- there is an attempt to resurrect the "CAD" group.
The Specialist Groups calendar contains highlights of activities.
iv. The Chairman has been in correspondence with Dr Barber (Vice-President,
Specialist Groups) on administrative support to our Group from headquarters.
v. There is a strong probability of another FORTRAN Forum being held in
London in July next year. A similar event took place in October 1981.
vi. Future meetings - on 10 December at BCS HQ, David Bailey (Salford
University) will talk on "Expert Systems from the FORTRAN point of view".
John Reid (Harwell) continues to attend X3J3 meetings and has kindly provided the
report in appendix C of the May meeting in Boston.
7. ANY OTHER BUSINESS
i. Recently our members were sent renewal notices which needed to be
completed and returned in order to remain in our Group. The returns have been
analysed by the Secretary to produce the up to date membership list in
ii. The Secretary has recently produced a CCTA Technical Note on the NCC/
FSTC FORTRAN 77 compiler validation scheme. The latest available certified
Compiler List, maintained jointly by FSTC and NCC, is in Appendix E.
8. DATE OF NEXT MEETING
The next meeting of the Group will be held OH Monday 19 September 1984
from 10.45 to 16.00 at BCS Headquarters. In the afternoon Geoff Millard of
Edinburgh Regional Computer Centre will give a talk on "The ICL FORTRAN 77 Optimising
Compiler".
9. THE ISO FORTRAN MEETING IN GENEVA
In the afternoon John Wilson gave a talk on the April meeting 0f the ISO FORTRAN
Working Group at CERN, Geneva. A summary of this appears in Appendix F.
MIKE NUNN
(Secretary)
25 JULY 1984
THE BRITISH COMPUTER SOCIETY
FORTRAN Branch/Specialist Groups
Receipts & Payments Account for the year ended 30th April, 1984
(Please submit to HQ by 3lst May)
Budget Actual
£ £p
Balance at start: Bank: Account 'A' 1421.47
Account 'B'
Cash:
Add Receipts: Cash received from HQ against Budget 350.00
Cost of HQ services: as notified
Cash received £rom HQ £or projects
Bank deposit interest received
Net income from Special Events: 82.00
Social Evenings, Conferences, etc
(See_separate statements attached) [1]
Net income from Branch Sub-Groups:
(See_separate statements attached)
Other income received (please specify)
Total: Balance + Receipts 1853.47
Less Payments: Meeting expenditure
Mailing expenditure
Secretarial expenditure
Committee expenditure
Professional services
Cost of HQ services received: 439.50
(as notified)
Project expenditure: 71.67
(See separate statements attached) [2]
Net expenditure on Special Events:
(See separate statements attached)
Net expenditure on Branch Sub-Groups:
(See separate statements attached)
Other expenditure (please specify)
Total Payments: 511.17
BALANCE AT END OF YEAR: (A) 1347.30
Made up of: Bank: Account 'A' 1347.30
Account 'B'
Cash:
Total Balance: (B) 1347.30
('A' and 'B' should agree)
Please enclose bank statement (or copy), and a list of outstanding
cheques.
Auditor's Report
I certify that the above Receipts and Payments Account for the
year................are in accordance with the books and vouchers of
the...................Branch, and that all the transactions carried out have
been proper to the purposes of the Branch.
.....................19.. Auditor...................
.....................Chairman/Secretary
T L van Raalte......Treasurer
[1] £50 from Academic Press for publicity; £2 each from 16 non-BCS members.
[2] Speaker's expenses £16.05; X3J3 Observer's fee £55.62.
SPECIALIST GROUP ANNUAL REPORT
YEAR ENDED 30 APRIL 1984
Name of Group FORTRAN
No. of Members 108
No. of BCS Members 86
No. and type of regular meetings 5 meetings in the year [2 at BCS Headquarters,
3 at Birkbeck College, London. Av. attendance 17 Format: whole day [Monday);
morning - business and language development/ standardisation work; afternoon -
seminar by invited speaker.
Studies undertaken
Publications Various articles & comments in trade press [COMPUTING, COMPUTER
WEEKLY etc]
Relations with other bodies Liaison with ANSI X3J3, BSI/OIS/5 and ISO/TC97/SC5/WG9
Other activities Delegation to ISO Fortran Experts meeting in Geneva April 1984.
Election of Officers Election for 1984/5 postponed to June, current officers:
Chairman: J.D. WILSON
Secretary: M. NUNN
Treasurer: T.L. van RAALTE
General Remarks
Projects for next year Maintain strong links with ISO and ANSI language
committees. Publicise, and monitor UK view oF draft standard Fortran 8X
due out in 1985.
Chairman J.D. WILSON
( signature )
Please return this form to Julia Allen, Liaison Executive (Specialist Groups)
The British Computer Society, 13 Mansfield Street, London WlM 0BP. Not
later than 27 April 1984.
To: BCS, NAG, DAP, etc.
From: John Reid
Date: 17 May 1984
Subject: Report on X3J3 meeting at Boston, 7-11 May 1984
References:
[l] 90(*)JCA-11 Approval of FIB-1 for public release
[2] Fortran 8x 'hit list'
[3] 87(*)JLW-4 Letter from Wilson to Wagener
[4] 88(*)JLW-2 Letter from Wagener to Wilson
[5] 90(.)JLW-1 Organisation of the Fortran 8x draft proposed standard
[6] 90(*)JCA-19 Fortran 8x survey
[7] 90(6)RAH-2 WHERE statements and transformational functions
[8] 90(13)AW/JKR-1b Rewrite of Chapter 14 of S7
[9] 90(7)KWH-1 CONDITION/ENABLE proposal
[10] 90(8)DDP-1 Treatment of initialized entities
[11] 90(9)LJO-1 Pointer proposal
[12] 90(9)JKR-6 Pointers
[13] 90(9)KWH-11 Pointers
[14] 90(9)BTS-2a Ranged integers
[15] 90(11)JHM-1 INQUIRE
[16] 90(11)JHM-4 PAD=
[17] 90(11)JHM-6 PROMPT=
[18] 90(9)JKR-3 Intrinsic for CPU time
Note: Formal votes are always two-way (Yes-No). Straw votes
are usually three-way (Yes-No-Undecided) but may
involve more choices.
l. Summary
There was much concern and discussion about the negative votes
of IBM and DEC [l] re publication of the Fortran Information Bulletin
(FIB). Since both were essentially saying that F8x is too big, the
steering committee held an evening meeting to consider reducing its
size and most X3J3 members attended. It was agreed that the language
is too large (15-4-4) and a 'hit list' [2] of 30 features that could
be removed was established. A straw vote on how many of these should
go was
more than 20: 3
10 to 19 : 8
3 to 9 : 8
0 to 2 : 3
and votes were taken on each feature(see[2]). It was concluded that
the committee would not agree to a sizeable reduction. DEC was also
concerned about the ease of transition from F77 to F8x and the
following extra wording for the FIB was agreed. "No Fortran 77
features will be removed; it remains X3J3's intent that any standard-
conforming Fortran 77 program will be a valid Fortran 8x program and
that, with exceptions clearly listed in the document, new Fortran 8x
features can be compatibly incorporated into such programs". This is
not a change of policy by X3J3; it merely spells the policy out more
clearly.
It remains the chairman's intention that fresh technical
proposals are confined to the August meeting and that polishing up the
document is the priority for the meetings from November onwards.
Individual members will be permitted to bring proposals to the
committee in August without going through subgroups (a "one-off"
dispensation). Straw voting on the whole document will begin at the
February meeting. Formal voting on releasing it for public comment
will be by mail ballot and it is hoped that this will happen later in
1985. Votes of "no" in the mail ballot will require explanations.
The committee has accepted my invitation to meet in Oxford, July
8-12, 1985. This will replace the August meeting in 1985.
2. Kenneth Wilson presentation
Kenneth Wilson (Nobel prize winner) corresponded with Jerry
Wagener last year, [3] and [4], essentially asking the committee to
stop work. He was invited to make a presentation to the committee at
this meeting. He was listened to with great attention and spoke about
the GIBBS project at Cornell. This aims to provide a software
framework for a scientist's work-station that captures the history of
the development of the work and modularizes it in a natural way (e.g.
for solving a partial differential equation, there would be separate
modules for the equation itself, the grid, the boundary conditions,
etc.). It would be a preprocessor that produces Fortran object code.
There was no mention of other similar work elsewhere and he did not
present a convincing case for an extremely conservative new standard.
3. Format of the draft proposed standard
Jerry Wagener [5] proposed a substantial restructuring of S7,
which currently is organised similarly to the F77 standard. His view
is that the new concepts (e.g. array features, derived data types)
make the old format inappropriate; after minor amendment his
proposal passed (24-0). It was not decided whether a new standing
document S8 would replace S7 or whether S7 could evolve to the new
format. Jerry undertook to write a key chapter (Chapter 2, Terms and
Concepts) for August and other committee members were assigned to
rewrite other chapters for November.
4. Adams questionnaire
Jeanne Adams prepared a questionnaire [6] to accompany the FIB
in the hope of getting reactions from users of the language, and asked
the committee to fill it in as a trial. I have added the results to
[6] because they give an indication of the sentiment of the committee,
but actual users are likely to have different views.
5. Array features
There is a problem with the use of [ and ] for array constructors
because these characters are not available everywhere. The motion to
allow (/ and /) as alternatives failed (11-12). A straw vote on
replacing [ and ] by (/ and /) failed (6-8-19) and one to replace them
by "something else" succeeded (14-7-6). A clean solution would be to
use < and > but this would mean that these characters could never be
used instead of .LT. and .GT. . A straw vote on this use of < and >
was undecided (11-9-10). It was hoped that someone would move such a
proposal in August.
There was an inconclusive discussion on the meaning of
"transformational" as an adjective for a user function. A proposal to
define a function to be transformational "if the result or any of the
arguments are array-valued" failed (7-13). The term is only used to
specify which functions may not appear in WHERE blocks and this was
resolved by changing "transformational" to "non-elemental".
Two proposals [7] for extending WHERE to transformational
functions were considered by the array subgroup. Making WHERE qualify
just the assignment was not liked (1-9-2) but allowing
transformational functions, with the interpretation that no WHERE
control is applied to their arguments, was liked (7-3-2) and was
passed by the main committee (13-8).
6. Rewrite of the definitions of the intrinsics
My work with Alan Wilson [8] rewriting the chapter of S7 that
defines the intrinsics, was well received. Jeremy Du Croz's
suggestion to order the full descriptions alphabetically was accepted
(27-3-1), but the wording "processor-dependent approximation to" for
functions (e.g. sin(x)) whose result is not exact was not liked (2-20-10).
With these changes and a few other minor ones, the rewrite was
accepted (25-0).
7. Event handling
Kurt Hirchert presented another tutorial on his approach [9] to
event handling, but the committee was undecided about its adoption
(6-6-25). It is very unfortunate that there is no collaboration between
Kurt and the EWICS group. It is particularly hard to understand why
they did not get together when Kurt went to Geneva. I feel extremely
sad at the prospect of seeing all their work, effort and expense in
attending meetings wasted but can see no prospect of the EWICS
proposal reaching a sufficiently polished state to be accepted in
August, the last meeting for technical proposals. I therefore
proposed a straw vote on the deletion of Chap. 19 of S7 and this was
favoured (16-2-7). A very fine presentation of the aims of the EWICS
group was given by Odd Pettersen, but he has not been strongly
involved with their Fortran proposal and will not be able to get to
the next meeting. His appeal that the proposal to delete Chap. 19
should not be moved at this meeting was respected.
8. Initialized entities
There are unresolved problems in relation to the interactions
between INITIAL and DATA, SAVE, and RECURSIVE. Dan Pearl's proposal
[10] that initialized entities be treated always as if SAVE were
specified was favoured (16-0-3).
9. Pointers
Three presentations on pointers were made. Dick Hendrickson
explained his proposal with Linda O'Gara [11] which involves the need
for a special character to access the pointer but no special character
for the pointee; my paper [12] supports this approach. Kurt Hirchert
explained a proposal [13] that does not need a special symbol and
decides by context whether it is the pointer or pointee that is
referenced. I put the case made to me by Mike Delves that
conventional pointers would be much simpler to incorporate at this
late stage in the evolution of the language and would be better
understood by everyone. A straw vote on wanting pointers was split
(15-17-6) and a straw vote on the different approaches went
11 Traditional (Delves)
2 Reverse notation (Hendrickson/O'Gara)
20 Contextual (Hirchert)
2 Undecided
9. Ranged integers
Brian Smith identified a number of problems [14] associated with
a proposal for ranged integers that parallels generalized precision
for reals. These are probably all solvable, but a straw vote did not
favour continuing work (4-16-6).
10. FORTRAN or Fortran
Discussion of the FIB revealed disagreement about the use of
capitals in the name of the language. Dick Weaver said that the
convention that is becoming established is that if a name is only
spoken as a sequence of letters (e.g. PL1) then capitals are used,
whereas if it is spoken as a word (e.g. Ada), then lower case is used.
This rule favours Fortran. A straw vote also favoured Fortran (23-7-4).
11. Input-output
Minor proposals for INQUIRE [15] were accepted (10-4, 16-0, 14-3).
Addition [16] of an extra optional parameter PAD to the OPEN
statement to permit padding of short formatted records by blanks was
accepted (16-0). The need for a means of supplying a prompt when
reading from a terminal was agreed (16-0-l) and Jim Matheny's proposal
was liked (8-6-3) but not put to a formal vote because it was not in
the pre-meeting distribution (I objected because I wanted to consult
Peter Kirby).
12. Intrinsic for CPU time
Opinion was mixed on my proposal for a CPU-time intrinsic.
Objections were based on really wanting the cost to be returned and on
the impossibility of defining the precise meaning, particularly in a
multi-processor environment. A straw vote on the general idea was
divided (7-7-2).
BCS FORTRAN SPECIALIST GROUP MEMBERSHIP
UK MEMBERS
|
A N S Addo |
J B Haseler |
T Sankey |
|
P Allatt |
J E Hickling |
D M Scales |
|
Malcolm J Appleford |
W S Hilder |
J L Schonfelder |
|
Kevin Ashley |
Dr I D Hill |
B H Shearing |
|
Elizabeth Aylmer-Kelly |
D J Holmes |
F B Smith |
|
Dr J C Baldwin |
Dr J S Hutton |
F J Smith |
|
Mrs D Balmer |
C R Jesshope |
W Swindells |
|
Professor D W Barron |
G J King |
A S Tak |
|
L F Bennett |
P M S Kraven |
M Tedd |
|
Mrs J A Bettess |
Chris Lazou |
R Thurston |
|
P D Bond |
Chris Little |
David M Vallance |
|
J Boyd |
William Little |
A J B Walker |
|
S G Brazier |
David Littlewood |
Howard K Watkins |
|
K Brown |
Peter Loftus |
Steve Watkins |
|
Mrs C J Cable |
D Lomas |
Andrew Westlake |
|
P H Cartwright |
C K Mackinnon |
Alan Wilson |
|
B J Cavalot |
C W Martin |
John Wilson |
|
U P Cheah |
Brian Meek |
|
|
R A Clement |
|
|
|
I P Cliff |
Dr Gantam Mitra |
|
|
I Cuthbert |
J D Murchland |
|
|
A C Davenhall |
David Muxworthy |
|
|
C Dellow |
David Ng |
|
|
J L Dyke |
Professor B Niblett |
|
|
John H Elkin |
K Normington |
|
|
D Fincham |
Mike Nunn |
|
|
William Flexner |
T D Palmer |
|
|
Dr B Ford |
D W Payne |
|
|
D P Fordred |
P Phillips |
|
|
A Georgiades |
Andrew Porter |
|
|
Mrs Anne Gosling |
C J Purchase |
|
|
Dr E W Gorczynski |
T L van Raalte |
|
|
Mrs J E Graves |
S M O'Regan |
|
|
A R Grant |
J K Reid |
|
|
Dr K Graupner |
L R Rice |
|
|
A Hall |
J Roberts-Jones |
|
|
D T Hall |
Dr R T Rowles |
|
|
Gary Harding |
G A Ruscoe |
|
|
D Harris |
M H Saeedi |
|
OVERSEAS MEMBERS
Frank Engel Jnr - USA
Ian M Hunter - Australia
Jeanne T Martin - USA
Dr Loren P Meissner - USA
M Metcalf - Switzerland
Dr M W Wigan - Australia
Dr Theodore J Williams - USA
SUMMARISED LIST OF CERTIFIED FORTRAN COMPILERS
The current list of certified FORTRAN compilers by the level for»
which they are certified is as follows:
Subset Level
Burroughs B20 FORTRAN Release 4.0
Conv. Tech. CT FORTRAN 9.0
DEC PDP-11 FORTRAN Version 5.0
DEC VAX PDP-11 FORTRAN Version 5.0
DEC DECsystem-20 Model 2020 FORTRAN-20 Version 7
DEC DECsystem-10 Model 1091 FORTRAN-10 Version 7
MODCOMP MODCOMP NASA Extended FORTRAN 78
Plexus Plexus FORTRAN77 Version 1.1
Full Level
Burroughs Large Systems FORTRAN77 Version 3.4.1
Burroughs Medium Systems FORT77 Version 6.7
CSC CSTS FORTRAN Release FTN148
CDC CDC Cyber FORTRAN5 Level 587
CDC CDC Cyber FORTRAN 200 (G201-25)
Conv. Tech. CTIX FORTRAN 77 Version 2.1
CRAY CRAY FORTRAN Translator (CFT)
D&B Computing IBM VS FORTRAN Release 3 Mod 0
Data General AOS/VS FORTRAN 77 Revision 2.21
Data General AOS FORTRAN 77 Revision 2.11
DEC VAX-11 FORTRAN Version 3.2
DEC DECsystem-20 2060 Model B FORTRAN-20 Version 10
Gould Gould FORTRAN-77+ Release 4.0
Gould Gould FORTRAN 77/X32 Release 2.0
Gould Gould PS-3000 FORTRAN 77 Version 2.1
Harris Harris H800 FORTRAN 77 Version 3.12 (SAU)
Harris Harris H800 FORTRAN 77 Version 3.12 (Non-SAU)
Honeywell DPS 8 FORTRAN (8)FZ3.0
Honeywell DPS 6 FORTRANA Release 2.1
Honeywell Multics FORTRAN Version 10.2
Honeywell DPS 8 FORTRAN-77 Release C00
IBM VS FORTRAN Release 3.0 (5748-F03)
ICL VME FORTRAN 77 Release 8003
Intel FORTRAN 86 Release 2.1
NCR VRX FORTRAN Version 25
Perkin-Elmer FORTRAN VII Release R05-01
Prime Prime FORTRAN 77 Rev. 19.4
Ryan-McFar. RM/FORTRAN Ver. 1.4.0
Salford Univ. FTN-77
Siemens FOR1 Version 1.51
Sperry Corp. 1100 Series ASCII FORTRAN Level 1OR1A
Stratus VOS FORTRAN Release 2.4a
Tandem FORTRAN T9202E01
United Info. CDC Cyber FORTRAN 5.1 Level 552
United Info. Cray CFT FORTRAN Version 1.10
Wang FORTRAN 77 Ver. 1.00.02
THE ISO FORTRAN WORKING GROUP MEETING AT CERN, GENEVA IN APRIL 1984 - REPORT
BY JOHN WILSON
1. Major emphasis: F77 - standardise existing extensions to
FORTRAN
F8X - make FORTRAN a modern, reliable,
portable language.
2. F8X milestones:
August 1984 - complete technical proposals
November 1984 - complete re-write of selected sections
February 1985 - complete detailed editing of the
language specification document S7
late summer 1985 - distribute draft standard for public
review
1987 - review public comments
1988 - publication of the new standard
X3.9 - 1988 ("FORTRAN 88")
3. F8X comprises:
----------------------
| | |
| Core | Deprecated|
| | features |
| | |
----------------------
4. A number of straw votes (39) were taken on the new language's content. Some
of the more interesting ones were :
F8X should contain deprecated features 21-0-3
Deprecated features can be edited out and leave a consistent document 22-2-0
Processors required to optionally flag use of deprecated features 10-2-11.
5. Jerrold Wagener (chairman, ForTec committee) wrote that among the additions
contemplated for F8X five stood out viz.-
array operations
improved facilities for numerical computation
programmer defined data types
facilities for modular data and procedure definitions
concept of deprecated features.
6. New facilities offered by 8X included:
i. Environmental inquiry intrinsics eg:
HUGE - largest number of argument type
EPSILON - very small positive number
ii. Event handling by user defined interrupts
iii. Free form statements
iv. Multiple statements per line delimited by semi-colon.
7. Deprecated features are features marked as deprecated in 8X and which may be
removed from 9X. They tend to fall into 5 categories viz:
i. storage association related
ii. facilities superseded by something better
iii. constructs considered had practice in current language theory.
Examples of deprecated features are:
- assumed size dummy arrays
- passing scalars to dummy arrays
- BLOCK DATA
- COMMON statement
- ENTRY statement
- EQUIVALENCE statement
- F77 source form
- statement functions
- arithmetic IF
- computed G0 TO (replaced by CASE statement)
- DATA statement
- DIMENSION statement
- DOUBLE PRECISION statement (replaced by PRECISION statements)
- F77 DO statement
8. Features still under development include:
exception handling
POINTER data type
BIT data type.