FORTRAN SPECIALIST GROUP

                     HELD AT 7 MANSFIELD MEWS, LONDON

                                        ON 20 MAY 1993

Present:         Chris Lazou                       HiPerCom Consultants

Mike Nunn                        CCTA

Ted Golton                         RAL

Tim Bartle                          Salford Software

Mike Geary                        NAG

Connie Peake                     University of Oxford

Miles Ellis                          University of Oxford

Neil Smith                          DRA Bedford

Lawrie Schonfelder          The University of Liverpool

1.        Welcome to the meeting

            The Chairman Chris Lazou welcomed everyone to the meeting.

2.        Apologies for absence

Apologies for absence were received from David Muxworthy, Dave

Vallance and David Harris.

3.         Minutes of the last Annual General Meeting

The minutes of the last Annual General Meeting were presented and


4.        Matter Arising

The proposal to spend less time on business and more time on technical

matters had still not been implemented.

5.        Chairman's Report

The Chairman welcomed everyone to this year's Annual General meeting.

The BCS Fortran Specialist Group met four times during 1992-93. We

used Blacknest for our September meeting where an interesting talk on

earthquake detections was given.

The November meeting took place at IBM Southbank with the afternoon

spent on learning about IBM's Fortran as well as HPF.

At the February meeting a presentation was given by Instrumatics on the

subject of reverse engineering tools.

In addition to tracking the X3J3 and WG5 activities on Fortran 90, Chris

Lazou gave a presentation on the current status of HPF.

We also sponsored a UNICOM seminar on Fortran 90, HPF and C in

scientific computing. A second UNICOM seminar on Fortran is scheduled

for 9 - 10 June 1993.

Under the auspices of HiPerCom Consultants Chris Lazou has written an

industry report which also includes a questionnaire on Fortran and


A one day workshop on HPF is to be held at the end of February or early

March 1994. This is to be a joint workshop with the BCS Parallel

Processing Specialist Group. Chris Lazou, T Lake, D Parkinson and A

Gupta are endeavouring to find a venue and speakers. A call for papers

will be going out with the next mailing.

The Chairman thanked members for their support during his tenure last


6.        Membership Secretary's report

Ted Golton reported that there are 171 members of which 111 are BCS.

Only 14 of the 60 other members had paid their membership fees.

It was suggested that too much money, time and effort is being spent on

sending mailings to people who are possibly not even interested in

receiving material.

After some discussion it was agreed to put the membership up to £10 per

year. It was also agreed that if members did not pay their £10 membership

fee by the dead line they would not receive any further mailings. BCS

members should also be asked if they wish to remain members of the

Fortran Specialist Group and if not they would also be deleted from the


7.        Treasurers report

The Treasurer presented his draft accounts. He guided the meeting

through the figures (Appendix A).  [The appendix is not available]

He reported that the VAT situation has changed and that the FSG will no

longer attract VAT for any transactions. The new VAT Registration

Number is GB618168724.

All previous sales invoices should be destroyed.

Regarding the charity situation he reported that from effect 1 January 1993

the Registration Charity Number is 292786 which is to be included on any

written or printed notice which requests money.

It was moved that we approach Martin Bowman to audit the FSG


8.        Election of officers

The following were nominated, agreed and seconded:

Chairman                Chris Lazou

Vice-Chairman        John Young

Secretary                  David Harris

Treasurer                  Ted Golton

Chris Lazou thanked the group for their support.

Secretary - Miles Ellis proposed and the group agreed that we approach

David Harris and put to him that the FSG will meet his expenses related to

his function as secretary up to a sum of £100 a year (4 FSG meetings).

9.        Minutes of the last FSG meeting held on 25 February

There were two minor corrections to the minutes:

-        Point 2 - Roger King should read Roger Keeling

-        Point 8 - last paragraph - Oxcom Computer Services Ltd should read

         αcom Computer Services Ltd.

The minutes were accepted as a true record.

10.        Report from X3J3 Representative (appended)

As received by email from Jerry Wagener.

11.        A Fortran Newsletter

Due to his busy work load Miles Ellis regrets that he is now unable to

commit himself to work on a Fortran Newsletter.

Lawrie Schonfelder suggested that the FSG collectively get some material

together and approach the BCS Bulletin to publish it. He also suggested

that we should also approach "Computing" which is linked with the BCS.

Miles has undertaken to contact both editors to see if they were willing to

take regular articles.


12.        A one day Forum on HPF

A one day forum on HPF will be held at the end of February or early

March 1994 (see attached). The venue is still to be decided. Any offers to

help Chris Lazou in the organisation of this forum will be gladly accepted.

[The forum information is not available.]

13.        BCS and any other business

I don't think there was any other business as I recall nothing worth noting


14.        Next meeting of the FSG

The next meeting of the FSG is scheduled for Thursday 16 September. It

was also agreed that the next AGM be held on 19 May 1994.

Informal JLW report of the 1993 May 10-14 X3J3 meeting

The principal objectives of this meeting were to (1) continue interpretation

processing as a high priority, (2) review with X3H5 the draft X3H5 model

binding to Fortran 90, and (3) prepare the initial X3J3 suggested

requirements input to the WG5 revision requirements process.

At the beginning of the meeting there were 19 interpretation requests in

X3J3/93-006r having a status of "X3J3 consideration in progress", 8 items

returned from the X3J3 letter ballot, 7 items returned from the WG5 letter

ballot, and one new request in the premeeting papers. At the end of the

meeting 11 of these 35 items have status "X3J3 consideration in progress",

17 have status "X3J3 draft response" and will be on the next X3J3 letter

ballot, and the rest are "X3J3 approved; ready for WG5". In addition, 8 new

requests were submitted at the meeting, all of which have status "X3J3

consideration in progress", making a total of 19 of the now 139 items in

X3J3/93-006r2 with this status. Thus the interpretation backlog at the end

of the meeting is the same as at the beginning of the meeting, with a

database that is 9 items larger.

The review of the X3H5 draft binding was completed, with a number of

suggestions given to, and accepted by, X3H5 for revision of the draft.

These include describing syntax and intrinsic procedures as in the Fortran

standard, revising the examples, and shortening the draft by making more

references to the model document and the Fortran standard (the binding is

intended to be a "thin" binding anyway, rather than a "thick" one which is

completely self-contained). In addition, X3J3 requests that the X3H5 model

document and the Fortran 90 binding to it be processed together, including

simultaneous public reviews.

The attention given to X3J3/93-004r, the database for suggested requirements,

took two principal forms. One was to review the status of the 18 items

originally in 004r and the other was addition of new items to the 004.

Because this was the last chance to add items before the WG5 meeting, a

flurry of activity resulted in 12 new items being added, all with status

"registered" (status 1) and no further discrimination at this time; there

will therefore be 30 items in 004r2, which will be submitted for the WG5

meeting. Of the original 18 items, virtually all having status 1, seven were

moved to status 6 (archival), meaning no further action recommended, and six

were moved to status 3 (recommended to WG5 for a future revision). Status 3

was further subdivided into 3a (recommended for the 1995 revision) and 3b

(recommended for the 2000 revision); items 4, 15, and 16 were placed

in 3a, items 12, 17, and 18 given 3b.

Jerry Wagener

1993 May 17

JLW informal report on the 1993 July 12-16 X3J3 meeting (#126) in Brussels, Belgium

The objectives of this meeting were to (1) continue processing requests for

interpretations, (2) respond to the WG5 Berchtesgaden resolutions, and (3)

begin planning for implementation of the 1995 requirements.

About two-thirds of the subgroup time was devoted to interpretation

processing. Two interpretation subgroups, PROC and nonPROC, divided the

backlog of 28 items, including five new items in the premeeting distribution

and four more submitted at the meeting. Of these, 12 were approved for

inclusion in the letter ballot to be issued next month. That leaves a backlog

of 16 requests (7,49,70,88,91,121,136/137,138,139,141,143,145,146,147,148,

149), plus whatever is returned from the letter ballot and any new requests;

these are assigned to a single subgroup responsible for all maintenance

activities in the near future.

The remaining subgroup time was devoted to organization of the technical work

related to the Fortran 95 requirements (WG5 Berchtesgaden resolution B9, WG5-

N930). The main result of this planning is X3J3/93-204r4, entitled RB9

Implementation PlanS, which is a chart containing 18 line items, one each for

B9 resolution items A1-A5, B1-B3, C1-C7 and three for item B4. Identified for

each 204r4 line item are: the responsible X3J3 subgroup, the principal X3J3

member contact, specific external contacts (if any), an external

"subcontractor" (if one), and specific deliverables for X3J3 meetings 127,

128, and 129. In a few cases the X3J3 member contact and appropriate external

contacts have not yet been identified. Of the 18 line items, two were

identified at this meeting for subcontracting - B4.2 (additional obsolescent

features, subcontractor BSI) and C2 (derived-type parameters, subcontractor

L. Schonfelder); in all likelihood, at the next meeting other items will be

identified for subcontracting. In most cases the meeting 127 deliverable is a

tutorial on that topic, possibly accompanied by a preliminary proposal. X3J3

has so far recommended that one topic (C7, derived-type I/O) not be included

in F95 - the rationale for this recommendation is contained in paper X3J3 /93-

231, which will be communicated to WG5.

Some attention was given to the editorial process for producing the 1995

revision. (Only six X3J3 meetings remain, meetings 127-132, before the draft

revised standard is scheduled for delivery to WG5.) Documents X3J3/93-229,

230 summarize the currently envisioned editorial process, which involves two

"specification" standing documents maintained by the editorial committee. One

of these, the 008, is for editorial specifications; the other, a new standing

document (009), will be used for specifications pertaining to the B9

resolution items and any other primarily technical items. The formats of the

008 and 009 are expected to evolve (as did that of the 006) as experience

indicates the most useful forms. Preliminary suggestions for 009 fields

include "status", "functional specifications", "implementation

considerations", "example applications", "detailed edits", and "history"; the

detailed edits field would be optional and only the functional specifications

would be subject to X3J3 approval of a 009 item. Meeting 128 and 129

deliverables in the B9 Implementation Plan (93-204) are largely "009

specifications".  A small drafting committee (membership as yet undetermined)

would have the responsibility of turning the F90 standard + 008

specifications + 009 specifications into the 007 draft revision.

X3J3, of course, would have to approve the 007 before submittal to

WG5 as the draft revised standard.

Finally, some adjustments were made to the X3J3 requirements process, to

bring the US process more in line with that used by other WG5 member bodies

and WG5 resolution B7. In the light of the WG5 requirements action, a cursory

pass through that part of the 004 not yet having received formal X3J3 action

identified one new item (0020, user-defined elemental functions) with near-

unanimous straw-vote support, and a couple of others with a majority in

favor. Major objectives at the next X3J3 meeting are likely to be refining

and finalizing the X3J3 requirements and editorial processes, as well as

executing the B9 implementation plan and continuing to provide Fortran 90


Jerry Wagener

1993 Ju1y 19