Minutes of the Meeting of the BCS Fortran Specialist Group

 held at BCS, 13 Mansfield Street, London on 28 November 1991


Present:        Mike Bennett      National Power

                Miles Ellis       Oxford University

                E Golton          RAL

                Carol Hewlett     LSE

                Chris Lazou       University of London

                David Muxworthy   University of Edinburgh

                Mike Nunn         CCTA

                M J Roth          AEA Technology

                Ravi Sikka        Ove Arup & Partners

                Tony Webster      Salford Software

                John Young        PE-MOD


1.      BUFFET LUNCH


        The afternoon meeting of the Group was preceded by a Buffet

Lunch organised by Mike Nunn. Copies of several books on Fortran

90 were on display as participants marked the completion of the

work on the current standard Fortran 90. It was agreed that a

contribution of £4 a head would cover the majority of the costs.

Mike Nunn was thanked for taking time to organise the lunch which

was enjoyed by all present.


2.      APOLOGIES FOR ABSENCE


        Apologies for absence were received from Ray Crawford, John

Reid, Lawrie Schonfelder and John Wilson.


3.      MINUTES OF THE PREVIOUS MEETING


        The minutes of the meeting of 16 May 1991 were still not

available. In the minutes of the 5 September 1991 the last

sentence at the end of the first paragraph on page 2 starting

"Also the US cannot ..." should be deleted. It was also noted

that the time on the Notice and Agenda for this meeting should

have read 12.30pm.


4.      MATTERS ARISING


        The Secretary, John Young, had drafted a letter to the

editor of the Computer Journal Algorithms Supplement concerning

the last issue which contained Fortran programs. The letter was

read to the meeting and approved.


5.      FINANCES


        This extra item was included at the request of the

Treasurer, E Golton, who had received several letters concerning

the BCS finances. The effect on the Group's accounts was that

we were being charged at 70p a cheque and £2.50 a month. The BCS

has given an assurance that all charges will be refunded. The

allowance for the year 1991/92 had been set at £500 which was

reasonable for a group like ours but not so for the larger

specialist groups. The Treasurer felt that there was now a real

need to define what sums of money were coming into and going out

of our account. It was noted that the membership reminders for


1991/92 had not been sent out.


6.      REPORT FROM X3J3 REPRESENTATIVE


        Miles Ellis reported on the recent X3J3 meeting which he

described as pleasant and busy in California. The meeting had

finalised 40 interpretations of the new Fortran 90 Standard, some

of them trivial but most were quite difficult. It was

interesting to note that most of the requests had come from

compiler writers.


        In the Standard itself several errors had already been

noted. In one instance the arguments had been reversed which is

easy to correct. But in other instances what was said was not

what was meant to be said. One "grey area" was in the INTERFACE

block definition. These obscure cases tend to be found when

writing compilers. However, the international standard is a

legal document so problems cannot always be easily sorted out.


        The X3J3 meeting had decided that the major part of its work

would be dealing with interpretations of the Fortran 90 Standard.

It was developing procedures to speed up the time taken to

produce interpretations and would also produce a standing

document. Currently the time taken was thought to be too long

but the matter was not quite resolved. The next meeting of X3J3

was in May 1992.


        The view of X3J3 was that there should be a revision of the

standard in 3-4 years. This revision would contain nothing new

and no new functionality but would be a tidying up exercise.

This new standard would be in addition to the maintenance

releases. X3J3 hoped that WG5 would accept this suggestion of

a new standard. Meanwhile WG5 had asked X3J3 to do the

interpretations of the Fortran 90 Standard.


        The US still thinks X3J3 should develop future Fortran with

X3J3 doing the technical development and WG5 deciding on the

content of the standard. There are 3 models for the future

development of Fortran: (a) all development done by WG5; (b) WG5

to determine various projects for a new standard; and (c) one

national body, ie X3J3, to be the technical committee with WG5

deciding the content of a new standard. After all, X3J3 has the

most expertise and previous standards were developed under ANSI

international standard rules.


        The Chair of X3J3, Jeanne Adams, was resigning having done

the job since 1977. There would appear to be a contest for the

chair between 2 people.


        The Fortran 90 Standard would be completing an ANSI ballot

soon with an expected 20-1 in favour (the 1 NO vote from Boeing)

with any re-ballot producing the same result. It had been

announced by a statement and a press release at the beginning of

the X3J3 meeting that the US DOD was not going to accept Fortran

90 but planned to stay with FORTRAN 77 as its standard. X3J3 did

not take the pronouncement too seriously and the DOD can be over-

ruled. It was also proposed that the suggestion be made that

Fortran 90 could be used as well as of ADA.


        X3J3 expected that Fortran 90 compilers would become

available during the next year.


7.      FORTRAN 90 QUESTIONNAIRE


        The Chairman, Mike Roth, had sent out a Questionnaire to

groups and firms following approval at the last meeting. He had

received three replies and combined them into a statement to be

given to the computer press. This statement is given as an

Appendix. With only three replies perhaps the questionnaire

should be sent to other organisations.


8.      A CONSTITUTION FOR THE GROUP


        The Chairman, Mike Roth, proposed that the Group should

consider adopting its own constitution. At present, specialist

groups use the BCS Constitution by default if they do not have

a formal one. The meeting agreed to consider a draft document

which should highlight the key differences with the BCS

Constitution and also give the rationale for the proposal. Mike

was asked to produce a draft constitution for the next meeting.


9.      ANY OTHER BUSINESS


        Dates of Meetings


        The dates of meetings in 1992 were proposed as follows:


        6 February 1992        'Data Structures in Fortran 90'

        14 May 1992            AGM

        3 September 1992       Visit to Blacknest Data Analysis Centre

                               'Using Fortran in Seismology'

        26 November 1992


        The venue for meetings in London is still uncertain.


        Problems at the BCS


        The meeting discussed several letters from the BCS which

highlighted the financial problems currently being experienced '

by the BCS. The meeting felt that as the Group was small the

effects would be relatively minor.


        Salford Software Marketing Ltd Mailing


        The Secretary, John Young, reported that the BCS had failed

to carry out the Group's instructions over the mailing of the

Salford Software items. John had written to the BCS but had yet

to receive a satisfactory reply.


10.      DISCUSSIONS ON FORTRAN 90


        There followed an informal discussion on Fortran 90.


John Young

Secretary

17 February 1992