Minutes of BCS Fortran Specialist Group Meeting held at
BCS HQ On 29 June 1989
Present: Arul Britto Cambridge University
David Butland Bradford University Software Services
Martin Counihan University of Southampton
Adam Dawson Bradford University Software Services
Miles Ellis University of Oxford
Dave Griffiths SSF
Mike Gunn University of Surrey
Carol Hewlett LSE
Peter Holland SSL
M Jackson Met Office College
Chris Lazou ULCC
Les Russell AWE
Paul White Met Office
John Wilson University of Leicester
Rick Woods University of Surrey
John Young PE-MOD
1. APOLOGIES FOR ABSENCE
Apologies for absence were received from Mike Nunn and David
Muxworthy.
The Chairman, John Wilson, apologised for the late start of
the meeting.
2. MINUTES OF THE LAST MEETING [20 April 1989]
The Secretary, John Young, apologised to the meeting for the
non-receipt by members of the minutes of the last meeting. This
was due to printing problems at BCS HQ exacerbated by the recent
British Rail strikes. (Secretary's note: On receipt of the
minutes the following members contacted me about the lateness of
posting - Alan Wilson, David Hill, and Brian Meek).
BCS Business
The Vice Chairman, Chris Lazou, corrected the item about the
Software Engineering Technical Committee in that it was not
winding up but considering its future role.
The Chairman, John Wilson, added that the Software
Engineering Specialist Group had formally closed (see item 6).
3. MATTERS ARISING
Scottish Sub-Group
There was nothing to report on the Scottish Sub-Group but
there had been problems in contacting the members in Scotland.
Financial Matters
The Treasurer, Miles Ellis, drew the meeting's attention to
both the Provisional Accounts for 1988-89 (Appendix A1 [of the April 1989
minutes]) which were presented at the last meeting and to the Final
Accounts for 1988-89 (Appendix A2).
The main difference was a further list of charges from BCS
on the Final Account which took the account over-budget. Even
though BCS HQ were slow in presenting these costs it was
considered that the Specialist Group's Executive was doing a good
job in difficult circumstances.
It was noted that the speaker for the January meeting was
not paid his expenses until June due to various problems within
the Group and at BCS HQ. (Secretary's note: A formal apology has
been sent to Graham Barber).
Miles Ellis then outlined his role as Treasurer of the
Group. The Group's account had now been centralised within the
BCS. For a small Group this centralisation creates more
paperwork but larger Specialist Groups may benefit. At the
present time VAT was still the responsibility of the Groups but
BCS still had to sort out the way it handled VAT. Generally,
Specialist Group Treasurers were being given more work by BCS
and, in particular, there was likely to be an increase in the
handling of VAT.
Miles Ellis asked the meeting to approve an "honorarium" for
a person, part-time, to look after the Group's VAT on a routine
basis. There then followed a general discussion of what incurred
VAT or not, particularly within educational establishments.
The motion was then placed before the meeting that the
Treasurer, Miles Ellis, will pay a small honorarium for
completing the Group's VAT return. The motion was approved with
the sum to be decided at the next meeting.
As there were no questions for the Treasurer, the Chairman,
John Wilson, thanked Miles for his work.
Dates of Meetings
The following dates were set with the probable subjects for
the afternoon sessions.
Thursday, 26 October, 1989 Expert Systems
Thursday, 25 January, 1990 Parallel Processing
Thursday, 5 April, 1990 AGM & 20th Birthday
Celebrations
Thursday, 28 June, 1990
There followed a discussion on the form of the 20th Birthday
Celebrations. It was agreed that the Group could subsidise the
cost of the celebrations. The meeting was undecided whether
there should be a luncheon, dinner, or even a separate venue.
Any Other Business
The Secretary, John Young, apologised for not yet writing
to X3J3 or to the Netherlands Fortran Group.
The X3J3 Representative, Miles Ellis, summarised John Reid's
report of the X3J3 meeting in New York.
[The report appended to these minutes is of the subsequent X3J3
meeting in Vienna.]
It had been a very busy meeting. X3J3 were obliged to reply
to every comment made during the last public review period.
These replies were finished during the week with a customised
response. There had also been a large number of editorial fixes
made to the new Draft Standard.
The problem with characters and multi-byte character sets
had yet to be fixed in the Standard. It had been decided that
the default set would be A-Z and that there would be a user
defined set. WG5 had suggested that all programming languages
should have facilities for multi-byte character sets. The use
of multi-byte characters in Fortran was not yet finalised but
probably would be during the next public review period of the
Standard.
Martin Counihan asked about control characters in character
strings. Fortran was sloppy in its definition of control
characters often treating them as ordinary characters. Also
there was a distinct difference between control characters and
graphics characters. The whole issue was "fuzzy-edged".
Miles Ellis then expanded on Bill Rinehuls role within ANSI
and his attendance at X3J3. He then commented on the ANSI/ISO
link; it was still possible that ISO will pass the current Draft
document and that ANSI will change this document and produce a
different Standard following the public review period.
Chris Lazou asked whether there were any major changes.
Miles Ellis replied that there were none but obviously both X3
and ISO could remedy any major alterations.
The Chairman, John Wilson, then announced that the X3 public
review period for the new Fortran Standard would be 4 months from
28 July to 24 November 1989. The ISO public review period had
yet to be announced but would be within the X3 time.
5. FORTRAN FORUM 1989
The meeting discussed the form of the 1989 Forum and decided
that there would be 2 venues - one in Central London and one in
Scotland probably Glasgow or Edinburgh. With the announcement
of the public review period for the new Fortran Standard it was
decided that the Forum would be held in the last week of
September. It was undecided whether BISL would be involved but
they would be approached. The most expensive item would be the
copying of the Standard document.
It was resolved to form a Working Group to organise the
Forum and would consist of John Wilson, Chris Lazou, Miles Ellis,
and Martin Counihan.
Miles Ellis would contact speakers for the Forum when he
attended the WG5 meeting at Ispra and the X3J3 meeting in Vienna.
(Secretary's note: See Forum details enclosed).
6. BCS BUSINESS
Specialist Groups Management Committee
The Chairman, John Wilson, reported on the recent meeting.
The following is a summary of the items covered.
(i) The Software Engineering Specialist Group had been
closed (or "laid down") even though they met twice a year
and had a membership of 600.
(ii) The following Specialist Groups have been formed:
(a) Safety-related Computer Systems
(b) Hardware
(c) Software Re-use
(iii) There was no further news on:
(a) the move of BCS HQ out of London
(b) Chartered Engineer status
(iv) There was a request for nominations for the grade of
Distinguished Fellow.
Technical Committee
It was reported that a proposal on standards for Quality
Management and Computer Software was a joint activity by BCS with
IEEE.
In the afternoon, the convenor of the Fortran on PC's sub-
group, Mike Gunn, presented his review of the Fortran compilers '
for PC's for which he had collected information during the past
year. These were Microsoft 4.1, LAHEY, WATCOM, SALFORD 386, and
PROSPERO F77. He said most Fortran compilers for PC's seemed to
be reasonable but to beware of those costing $100.
Mike Gunn then introduced the following who spoke on their
experiences with Fortran on PC's:- Rick Woods on Microsoft 4.1,
Carol Hewlett, Arul Britto with Micro-CRISP, David Butland with
Simpleplot, and Miles Ellis on using Prospero Fortran for his new
book. It is hoped to include summaries of these talks as
Appendices in a later mailing.
John Young, Secretary
17 August 1989
To: Fortran Forum, BCS, NAG, etc.
From: John Reid
Date: 21st August 1989
Subject: X3J3 meeting in Vienna
Note: This is a personal note on the meeting and in no sense
does it constitute an official record of it.
1. Summary
This was a gentle meeting, a welcome contrast to the two previous
meetings. There were no late-night sessions (even of subgroups), it
was possible to read all the documents as they appeared, and we
adjourned on Friday morning. The meeting was not as well-attended as
usual, but a majority of the membership was present allowing
corrections to S8 to be approved. These corrections will be
accumulated in a new standing document, S18.
2. Ispra meeting of WG5
The ISO Fortran Working Group met during the previous week at
Ispra. Its principal conclusions were:
1. Acceptance of S8 as the second draft proposed standard DPl539.
2. Recommendation that the ISO ballot close on January 19 to permit
X3J3 to receive comments from the public and form a recommendation on
the US vote at the January meeting of X3J3.
3. Despite a wish for technical changes, a strong request to X3J3 to
limit changes to the correction of errors and inconsistencies, in the
interest of timely adoption of the new Fortran standard.
4. An intention that DP1539 become the sole international Fortran
standard and that there be no subsets.
5. A request to establish an ISO project for the development of a
varying-character module as a collateral standard.
6. A request to X3J3 to remove the feature that allows arbitrary
character sets in identifiers, to avoid possible incompatibility with
other standards under development.
3. Fortran 77 as a subset or a separate standard
Bill Leonard (Harris) wrote directly to X3 requesting that Fortran 77
be retained as a collateral standard, and this was considered by SPARC
(Standards Planning and Requirements Committee) in the week prior to
the X3J3 meeting. X3J3 was not consulted by SPARC nor formally told of
its conclusion, which was a recommendation by a vote of 13-1 to
continue Fortran 77 (without changes) as a standard and change Fortran
8x to becoming a new standard rather than a revision. X3J3 did not
understand the implications of this recommendation and it effectively
dampened discussion of subsetting. Jeanne Adams will make
enquiries. She thought that it might not be a big issue - the
intention may be to continue Fortran 77 for a little while and then
let it die. I kept quiet because this is really American business, but
it seems totally at variance with last year's directive from X3 that
the US support a single Fortran standard.
4. Arbitrary character sets in identifiers
A straw vote (21-5-5) favoured removal of the feature that allows
arbitrary characters in identifiers, but no formal motion was made
because no such proposal was in the pre-meeting distribution and it
was felt that more members should participate in such a decision. The
Japanese observer said that that there was no support for this feature
in the Japanese Fortran Group. At the WG5 meeting, there was little
support from the countries that might be expected to benefit.
5. Namelist data
An ambiguity in NAMELIST data was identified, illustrated by the record
&WHAT A= AA BB B= cc /
and the declarations
CHARACTER(LEN=2) A(4),B
NAMELIST /WHAT/ A,B
Does the record give 4 values to A or 2 to A and one to B? Andy
Johnson suggested requiring character values to be delimited and this
was favoured (18-4-11) but no formal proposal was made because it was
felt that further consideration was needed.
6. Public comment period
The ANSI public comment period runs from 27 July to 24 November.
Copies of S8 will be available from
Global Engineering,
2805 McGaw,
Irvine,
Ca 92714,
U.S.A.
Phones: (800)854-7179 (US), (714)261-1455 (International),
(714)261-7892 (Fax).
Copies will be available within the UK from the BCS Fortran Group at a
price of about £15.
The November meeting of X3J3 has been postponed to January 7-12 to
allow time for the comments to be distributed to members
beforehand. ISO WG5 has recommended that the ISO ballot close on
January l9 to allow X3J3 to decide on the US vote at its January
meeting. It is worrying that some X3J3 members are determined to use
the comments as a kind of opinion poll, giving equal weight to a
pro-forma letter as to a carefully researched opinion from a group of
people. There is little we non-Americans can do to counter this ~ our
letters will probably (and rightly) be excluded when forming the US
vote. Nevertheless. X3J3 remains the technical committee that has
been delegated the responsibility for preparing the Standard, so do
write in with your opinion and in particular let the Committee know of
any errors or technical flaws that you find. The address is
Fortran Public Review - attention Lynn Barra,
X3 Secretariat / CBEMA,
311 First St N.W., Suite 500,
Washington DC, 20001-2178,
U.S.A.
The UK vote in the ISO ballot will be made on the recommendation of a
BS1 panel chaired by David Muxworthy. The BSC Fortran Group will be
holding a Forum in London on 27 September and in Edinburgh either the
day before or the day after. Opinions expressed at these meetings will
be taken into account by the BSI panel.
7. Future X3J3 meetings
It was decided that the next meeting should be in October and should
formally be considered as a continuation of the Vienna meeting, but it
was later changed to an ad hoc meeting following an objection by Dick
Weaver (IBM). The meeting will be in Boulder, Colorado, October
16-20. It is needed in order to work on errors in S8 prior to the
January meeting which will be dominated by consideration of public
comments. The next regular meeting will be Richardson, Texas, January
7-12.
8. Metcalf and Reid
Mike Metcalf and I have revised our book to correspond to the second
draft standard and have taken the opportunity to add a few more
explanations, examples, and exercises. It is a little longer than the
first edition and will be published in UK on 7 September in paper-back
at £12.50 and in USA later in September at $22.50.