Minutes of the Meeting of the BCS Fortran
Specialist Group held at the Blacknest Data Analysis Centre,
Brimpton, Near Reading, Berkshire on 3 September 1992
Present: Tim Bartle Salford Software
Miles Ellis Oxford University
E Golton RAL
Mike Gunn Surrey University
Gary Halsall NAG
Peter Holland SSL
Chris Lazou HiPerCom-Consultants
Mike Nunn CCTA
Jonathon Wheeler RAL
John Wilson Leicester University
John Young PE-MOD
1 WELCOME
The Chairman, Chris Lazou, welcomed everyone to the meeting
held in the Library at Blacknest. The Vice-Chairman, John Young,
also welcomed everyone and gave the notices. Gary Halsall,
attending for the first time, suggested members introduced
themselves which duly took place.
2 APOLOGIES FOR ABSENCE
Apologies for absence were received from Mike Geary, Les
Russell and Dave Vallance.
3 MINUTES OF THE MEETING OF 14.5.92
The minutes of the meeting of 14 May 1992 were approved.
One correction was made under item 5 Matters Arising in that Jim
Bartle should be Tim.
The Secretary, John Young, offered profuse apologies to
those people at NAG Ltd, Robert Iles and Gary Halsall, for a
complete mix-up of names in the notice of this meeting sent with
the minutes.
4 MATTERS ARISING
It was thought that the cost of a copy of the ISO Fortran
90 Standard from the BSI was closer to £160 rather than £80.
In the US, there had been a procedural hold-up in the ANSI
Fortran 90 Standard. It was thought that the announcement would
be made during the recent X3J3 meeting but it was not. The
announcement could still be in a month or two.
Three members had "manned the stall" for the Group at the
Software Development 92 Show at Wembley. The Group thanked Chris
Lazou, John Young and Tim Bartle for their time. It was noted
that BCS was not "visible" in the User Group Lounge or at the
Show.
5 DATES OF FUTURE MEETINGS
It was agreed that the next meeting would be 26 November
1992 at IBM Southbank but not necessarily with speakers from IBM.
The first meeting of 1993 was confirmed as 25 February 1993
and the Group would arrange to meet in the BCS London venue in
7 Mansfield Mews.
After some discussion of dates of X3J3 and other meetings
it was agreed to delay the AGM by one week from 13 May 1993 to
20 May 1993. No venue for this meeting has been arranged.
6 MEMBERSHIP LIST
The Treasurer, E Golton, is now Membership Secretary and
reported that the Group had 2 new members. The current
Membership List is woefully out of date and Ted was asked to
contact BCS HQ to organise updating. If this task had not yet
been started the Group needed to know a completion date.
The report on the WG5 meeting in Victoria, BC, was presented
by Miles Ellis and is included as Appendix A. The Standing
Document 4 was re-written by the meeting. A revision of the
current standard Fortran 90 should be produced in about 5 years.
There was also a call for suggestions for new Fortran (or
possible extensions) from participating countries.
The X3J3 Representative, Miles Ellis, then reported on the
X3J3 meeting in Seattle, WA. Appendix A also includes this
report. There was a general consensus accepting the Standing
Document 4 but neither ANSI or X3 had yet agreed but probably
would. The major item of work was the continuing processing of
"interpretations" of the Fortran 90 Standard. There was also an
upsurge in extensions to Fortran in bindings, etc.
9 FUTURE DIRECTION OF THE GROUP
Miles Ellis presented a discussion paper on the future of
the BCS Fortran Specialist Group (included as Appendix B). Miles
led the discussion by pointing out that item 2 was the strategy,
item 3 was the mechanics, and item 4 was a proposal for a new
publication. On item 2 in point two the phrase "the main" was
to be changed to "a main" and the items on the BSI Panel were to
be merged. The meeting discussed the paper and included the
following suggestions; (a) Applications of Fortran and (b) Look
at times/venues of meetings as a whole. It was agreed that the
discussion paper was a worthwhile "wish-list" and should be
presented to members of the Group as a whole. Miles' conclusions
were endorsed by the meeting.
10 BCS AND ANY OTHER BUSINESS
UNICOM Seminars
UNICOM Seminars asked the Group for separate mailing to
members to promote its forthcoming Seminar "Fortran and C in
Scientific Computing" with the Group's Chairman, Chris Lazou, as
the Seminar Leader. After some discussion, it was agreed that
as the Group were co-sponsoring the Seminar this mailing would
be done at cost by the BCS. Chris agreed to provide an
appropriate covering letter.
Treasurer's Report
The Treasurer, E Golton, presented a brief financial report.
The Group's allowance for this year was £500 and apparently last
year's shortfall had been written off by the BCS.
Gift in Appreciation
The Vice-Chairman, John Young, suggested that the Group
donate a small gift to the Blacknest Secretary, June Bojie, who
had typed all the Group's Minutes over the past four years. The
meeting authorised the sum of £10 to be used as John thought fit.
11 THE BLACKNEST DATA ANALYSIS CENTRE
The Centre's Leader, Professor Alan Douglas, was thanked by
the Chairman, Chris Lazou, for inviting the Group to Blacknest.
He also thanked the retiring Secretary, John Young, for his
contribution to the Group. Professor Douglas then talked about
the work at Blacknest before introducing Peter Marshall, OBE, who
gave a brief history of Blacknest with particular reference to
the work with the UN Conference on Disarmament in Geneva.
Peter's review was followed by a talk given by Hilary Trodd on
the current work at Geneva. Finally, John Young gave a brief
review of his work for the Group over the last 30 years.
12 THE NAG FORTRAN 90 COMPILER
Gary Halsall gave a brief talk on the NAG Fortran 90
Compiler and afterwards demonstrated the compiler working on the
Blacknest VAX Computer.
13 TOUR OF BLACKNEST
The meeting concluded with an optional tour of the Blacknest
facilities.
John Young, Acting Secretary
27 October 1992
Report on recent WG5 and X3J3 activities
WG5 and X3J3 held back-to-back meetings in the Pacific North-West during late July
and early August.
The WG5 meeting in Victoria, BC, was held first, from 27th to 31st July, and was
attended by 32 members, of whom 17, including the Convenor, came from the United
States; the UK, with 6 members (David Muxworthy, Malcolm Cohen, Miles Ellis,
John Reid, Mike Roth and Lawrie Schonfelder), was the next largest delegation, and
other delegations represented Canada, the host, Japan and Germany, with two
members (Aurelio Pollicini and Dick Hendrickson) having no formal national
accreditation.
The meeting concentrated on two main issues - "interpretations" of the 1991 Standard
and procedures for the future development of Fortran. In accordance with ISO
Directives, and with normal practice, the decisions of the meeting are recorded in the
form of Resolutions, a copy of which is appended to this report. A key document,
referred to in Resolution V6, is the Strategic Plan for Fortran Standardization (WG5
Standing Document 4), which is also appended to this report.
There was a considerable debate over the choice of a Primary Development Body for
the next revision, and indeed over the relationship between that body and any other
development bodies. In the end it was agreed that X3J3 should be asked to become
the primary development body as long as the work was carried out as an X3 I-Project and
that the relevant I-Project was established by june 30th next year (see Resolution V9). This
decision was endorsed by the SC22 Plenary Meeting in Tampere, Finland, last week,
which asked the US member body to take all necessary steps to ensure that the
appropriate I-Project was so established.
The X3J3 meeting (122), the following week, was held in Seattle, WA. It spent a
significant amount of time discussing the Strategic Plan approved the previous week
by WG5 and an attempt was made to demand a much higher degree of autonomy
from WG5 than WG5 had indicated that it was willing to allow. In the end, the
proposal (initiated from the chair) was voted down by a substantial majority.
The major item on X3J3's agenda was the continuing processing of "interpretations"
and it is clear that, both procedurally and administratively, the present procedures
need streamlining if the committee is not to drown under the weight of this work.
The other important issue concerned relationships with other bodies who are
proposing to develop bindings and/or extensions to Fortran. Specific examples
which were discussed are POSIX, X3H5 (Parallel Computing) and HPFF (High
Performance Fortran Forum).
Miles Ellis
2 September 1992
The FSG minutes here incorporated the WG5 Resolutions,
which are available at
https://wg5-fortran.org/N001-N1100/N814.pdf.
ISO/IEC JTC1/SC22/WG5 - N820a
Strategic Plan for Fortran Standardization
(WG5 Standing Document 4)
1. Mission (the goal to achieve)
The mission is to be responsive to changing technology and user needs by
establishing procedures whereby a revision of the international Fortran
standard is produced about every five years. Each of these revisions, during
its lifetime, will be recognized and accepted worldwide as the single
international Fortran standard.
The standard will be developed and maintained by a broad and balanced
international group of users, vendors, and academics, collectively providing
appropriate expertise.
2. Vision (the desired effect of achieving the goal)
The international scientific and engineering community will be provided with
a single worldwide Fortran standard that will improve software and programmer
portability and promote software reusability and compatibility.
3. Strategies (what specific functions will achieve the goal)
3.1. Two major functions
There are two major functions: (1) identification of needs and specification
of standard requirements and (2) development of standard document content and
maintenance of the current standard. Other activities include scheduling,
performing liaison and review activities, and conducting ad hoc studies. The
requirements and development functions should be separate but highly
coordinated.
3.2. Requirements as the "what" function.
The result of requirements specification is a document that prescribes the
functional requirements for a revision of the language standard and related
standards. Draft documents (standards, amendments, corrigenda, technical
reports, etc.) will be submitted to ISO for adoption.
WG5, on a continuing basis, determines, records, distributes, and maintains
the needs and suggested requirements for Fortran. To begin a new revision
WG5, using the recorded needs and suggested requirements, establishes
objectives and corresponding functional requirements, specifications, and
schedule for that revision.
3.3. Development as the "how" function
The development function develops the international standard revisions,
corrigenda, amendments, and related auxiliary standards, in accordance with
the specified requirements and schedule.
A development body will be chosen to produce each revision and auxiliary
standard. The development body for a revision of the Fortran standard is the
primary development body. It is the responsibility of each development body
to determine how the requirements are to be met and to prepare the
corresponding draft document. Revisions of the Fortran standard will
incorporate approved corrections and clarifications as well as fully
implemented requirements.
Maintenance of a standard involves correcting errors that are discovered in
the standard and providing official interpretations for those parts of the
standard that are found to be unclear or incompletely specified.
Corrigenda that incorporate corrections, clarifications and interpretations
resulting from this maintenance activity will be produced periodically.
3.4. Scheduling.
The basic five-year schedule for revisions is predetermined. NG5 will
negotiate with the primary development body a detailed schedule consistent
with this five-year cycle. For other development work WG5 will negotiate an
appropriate schedule with the corresponding development body. Activities
involving maintenance of the current standard will take place concurrently
with the development activities.
Since the schedule must be reported to JTC1, the schedule will be phrased in
ISO terms. The relevant ISO stages for ongoing projects are:
2 preparatory
3 committee CD ballot (SC22)
4 approval DIS ballot (JTCl)
5 publication
[See, for example, WG5 Standing Document 3 (WG5-N643)]
3.5. Auxiliary standards.
Some needs may be met most effectively by auxiliary standards rather than with
revision of the Fortran standard. An auxiliary standard is related to the
Fortran standard in some significant way; for example, some functionality may
be made available by "binding" that functionality to Fortran. There are two
forms of auxiliary standards: supplementary standards and extension standards.
Extension standards involve syntax/semantics beyond that of the current
Fortran standard. Because of the serious potential for incompatibilities
arising from uncoordinated extension activities, WG5 requests that all bodies
developing extension standards coordinate their work closely with the primary
development body and that primary development body approval be obtained before
a proposed extension standard is adopted.
Supplementary standards are auxiliary standards other than extension
standards. Common forms of supplementary standards likely will be standard
Fortran code using the module/use and procedure library mechanisms. In order
to minimize conflicts and incompatibilities, WG5 requests that the primary
development body review all proposed supplementary standards and provide WG5
with the results of those reviews.
3.6. Liaison and review activities.
It is the responsibility of WG5 and the development bodies to establish
liaison or contact with and review the work of other organizations whose work
items relate to the Fortran standards effort. Such organizations include:
SC22/WGl1 and X3T2 (CLIA, CLID, CLIP)
SC22/W615 and IEBE P1003.9 (POSIX Fortran Bindings)
SC22/WG20 (Internationalization)
X3H5 and HPFF (Parallel Processing)
3.7. Conducting ad hoc studies.
It is the responsibility of WG5 to determine whether studies are needed to aid
the need identification and requirements specification processes. WG5 may
carry out such ad hoc studies or may request one or more participating member
bodies to carry out such studies.
4. Tactics (how to achieve the strategic objectives)
4.1. Principal responsibilities of W65.
WG5 may establish a management committee to make recommendations and interim
decisions when it is impractical to involve the entire membership. The
membership of the management committee would be reestablished at each meeting
and be in effect until the following meeting.
W65 will:
(a) maintain a set of needs and suggested requirements
(b) define revision objectives and select the corresponding functional
requirements for development after comprehensive review
(c) determine whether a given requirement can best be met by a revision
of the Fortran standard or by an auxiliary standard (see N776)
(d) choose the development bodies and delegate all development and draft
document preparation to them
(e) monitor development (where progress is not consistent with the
schedule, WG5 will, together with the development body, determine
if corrective action can be taken by the development body; if this
is not possible it will be necessary for WG5 to reassign some or
all items to another development body, or to delete some items
from the specification, or to revise the schedule)
(f) monitor maintenance activities
(g) determine whether a final draft document is adequate and consistent
with the specified requirements
(h) submit draft documents to ISO
4.2. Principal responsibilities of the development bodies.
A development body will:
(a) appoint a project editor
(b) produce a draft document to meet the agreed schedule and
specified requirements
(c) establish a document production system that will allow the most
recent electronic version of any proposed standard to be available
to all members of WG5 and the development body
(d) coordinate with other organizations, as may be appropriate
4.3. The schedule.
Following is a proposed schedule for the next two revisions:
phase first revision second revision
preparatory 2.1 11-91 7-92
initial requirements specified ---- 7-94
requirements review complete ---- 5-95
requirements specified completely 7-94 9-95
working draft 2.8 4-95 12-98
CD ballot 3.0 6-95 1-99
3.1 6-95 1-99
3.2 9-95 4-99
3.8 11-95 9-99
DIS ballot 4.0 12-95 10-99
4.1 12-95 10-99
4.2 7-96 5-00
4.4 9-96 7-00
4.8 9-96 7-00
publication 5.0 11-96 8-00
5.3 12-96 10-00
4.4. Fortran maintenance bodies.
WG5 will determine maintenance bodies for Fortran revision standards and
auxiliary standards. In general, the body that develops a standard will be
responsible for its maintenance. X3J3 has been delegated responsibility for
maintaining Fortran 90.
WG5 will conduct a letter ballot on items that the maintenance body identifies
as ready to be included in a given corrigendum. WG5 will forward the
resulting draft of approved items to SC22 for further processing.
4.5. Coordination between WG5 and development bodies.
Meeting schedules will be coordinated to make the best advantage of members'
resources as far as time and travel expenses are concerned and to accomplish
the goals established by the schedule.
There should be as much overlap as possible in the membership of WG5 and
development bodies, especially with the primary development body. This
provides informal communication and continuity.
Formal written reports and resolutions will be used to clarify communications
and maintain a record of communications. At each meeting a Responses and
Resolutions (R&R) report will be produced that is intended to be the official
communication and coordination vehicle between WG5 and the development bodies,
as appropriate. The R&R reports produced by WG5 will be sent to the
appropriate development body; the R&R reports produced by the development
body will be sent to WG5. The responses portion of each R&R report will
contain responses to resolutions received from the other bodies. The
resolutions part may contain any information for or requests of the other
bodies.
Annex A. Schedule Implications
For the next three years, the proposed schedule has the following
implications for each participating WG5 member body:
before - review the strategic plan
1993 - review the maintenance activities of X3J3
meeting - prepare suggested requirements accompanied by
rationale and send to Convenor at least 6 weeks
before the meeting
at - adjust the strategic plan
1993 - draft recommendations for Fortran 90 maintenance
meeting - review suggested requirements, create
repository of needs and suggested requirements
before - prepare informal list of requirements accompanied
1994 by rationale and send to Convenor at least 6
meeting weeks before the meeting
- review ongoing maintenance activities
at - finalize requirements for revision
1994 - draft recommendations for Fortran 90 maintenance
meeting
before - review WD for revision
1995
meeting
(spring)
at - approve WD of 1996 revision for CD registration
1995
meeting
(spring)
before - prepare national responses to CD ballot
1995 - prepare list of requirements and
meeting send to Convenor at least 6 weeks before meeting
(autumn)
at - process results of CD ballot on revision
1995 and approve edits to draft for DIS registration
meeting - finalize requirements for 2000 revision
(autumn)
before - prepare national responses to DIS ballot
1996 - review development activities
meeting
at - process results of DIS ballot and approve edits
1996 to draft for publication
meeting - monitor development of requirements
The Future of the BCS Fortran Specialist Group
A Discussion Paper
1. Background
In recent years the main activity of the FSG has been monitoring the development of
Fortran 90 and coordinating reaction from the UK Fortran community. With the
publication of the ISO Fortran 90 Standard last year it is time for the Group to
reassess its role and establish new goals and/or activities. This paper does not
attempt to discuss any of the possible new directions in any detail; instead it is
intended to be more of an initial check-list of possible options for discussion by the
Group as a whole.
Section 2 of this paper lists some of the major objectives which the Group might
consider adopting in the future - the strategy. Section 3 then lists some of the possible
ways in which these objectives might be attacked - the tactics. Finally, section 4
makes a proposal for one of the tactics detailed in section 3.
2. Possible strategic objectives for the FSG
3. Possible tactics for the FSG in order to achieve its objectives
4. A Proposal for a Fortran Magazine
In the United States there are currently two newsletter/magazines devoted solely to
Fortran. Both are quite widely distributed and are available only by subscription.
The first of these, Fortran Forum, is published by the ACM (through SIGPLAN I
believe) and is edited by Loren Meissner at the University of California. There are
two or three issues each year at somewhat random intervals, although it is reported
that the ACM has demanded a more regular publication schedule in future.
The second, and more recent publication, Fortran Journal, is published every two
months by the Fortran Users Group and is edited by Walt Brainerd. It appears to be
funded entirely by subscriptions and advertising, although it is not clear whether the
Fortran Users Group has any role in life other than as the recipients of the Fortran
journal!
I believe that there is scope for a similar newsletter/magazine in this country, and I
would be interested in editing such a magazine. I propose the name Fortran Matters
for this publication which would be both a vehicle for publicising the activities of the
FSG and a means of generating more interest in, and information about, Fortran in
general.
I am sure that the advent of Fortran 90 provides the opportunity for Fortran to
recapture much of the ground that it has lost in recent years through being,
undeniably, old-fashioned and lacking in features. However it is a larger and more
sophisticated language than Fortran 77 and so could easily drown under its own
weight if it is not publicised properly. Even on the conservative western side of the
Atlantic, though, there are increasing noises about how much the programming
community has to gain from Fortran 90, and it has even been reported that a number
of U.S. Computer Science Professors are seriously considering moving to Fortran 90
as the initial language for Computing Science undergraduates!
If the Group wishes to pursue this idea I will bring a formal proposal to the next
meeting with costings and an indication of the likely subsidy required to establish the
magazine, together with, hopefully, an indication as to the possible sources of
financial support. There are two possible variations on the same theme:
i) Fortran Matters would be published by the BCS (or by the BCS FSG),
which would therefore carry the ultimate financial responsibility;
ii) Fortran Matters would be published by an external organisation, in
association with the BCS, with the external organisation carrying the
ultimate financial responsibility; I have a possible publisher in mind.
In the latter case it is assumed that the BCS (or the BCS FSG) would provide some
form of financial subsidy, but its financial commitment would be strictly limited - as
would its editorial influence. In the present circumstances, however, I imagine that
the second option is the only realistic one.
5. Conclusions
Unless some decisive action is taken soon the membership of the Group may dwindle
to the point where it would be pointless even to try to revive it. We need to take
action now, or accept that the era of the BCS Fortran Specialist Group is at an end.
Miles Ellis
2 September 1992