BRITISH COMPUTER SOCIETY FORTRAN SPECIALIST GROUP



Minutes of a meeting held on

Friday, 3 May 1974 at the

Royal Society of Tropical

Hygiene & Medicine, 26 Portland

Place, London W1 at 10.30 am.



Present:    Mr B.H. Shearing (Chairman)     Alcock Shearing & Partners

Mr D. Abel                      University of Edinburgh &

                                University of North Queensland

Mrs E. Aylmer-Kelly             University of York

Mr B.J. Banes                   Rolls Royce, Bristol Engine Division

Mr E.O. Bodger                  IBM Systems Aids Centre

Mr P.D. Bond                    Philips Industries

Mr H.W. Bradly                  CAD Centre, Cambridge

Mr P.A. Clarke                  Rothamsted Experimental Station

Mr J.C. Cullen                  B.P.

Dr A.C. Day                     University College, London

Mr D. Ellison                   University of Manchester R.C.C.

Mr M. Fitzsimmons

Mr G. France                    C.E.G.B.

Mr P. Hartley                   Honeywell I.S. Ltd

Mr I.D. Hill                    MRC Computer Unit

Mr J.P. Holland                 Tunnel Cement Ltd

Mr M. Kerr                      C.E.G.B.

Mr M.J. King                    B.B.C.

Mr D.J. Maisey                  I.C.L.

Mr T.L. van Raalte              A.W.R.E. Aldermaston

Mr I. Williams                  Kleinwort, Benson Ltd

Mr D. Winstanley                University of Birmingham

Mr D.T. Muxworthy (Secretary)   University of Edinburgh


Apologies

for         Mr J.S. Gatehouse                I.C.L.

Absence:    Mr T.D. Palmer                   Computer Power

Mr K. St.Pier                    G.E.C.

Mr A.J.H. Walter                 Atlas Computer Laboratory


1.    APPROVAL OF              The minutes of the meeting of 15 March 1974 were     ACTION

MINUTES                  approved subject to two corrections: (1) at section

2b. Mr Hill's letter was not on behalf of the Group,

(2) the second paragraph in section 5 should have

read, "Clarification of the position of the BCS

Fortran representative on the BSI Programming

Languages Committee . . . . . .".

 


2.    MATTERS ARISING          a.    It had been learned from the BCS that for the

      FROM THE MINUTES               Society to produce 200 copies of the proposed

draft American National Standard for Fortran

would cost £194. It had been suggested that

the cooperation of the BCS Standards Committee   BHS

be sought in this matter. There was no

indication yet of ANSI's official attitude.


b.    There was some discussion of the article in

"Computer Weekly" (18 April 1974 pp 6-7) and a

reader's reply (2 May 1974 p 2) concerned with

character manipulation in present Fortran. It

was decided that the article did not give

grounds for formal action by the Group and it

was left to any individual member who wished to

reply to point out that provision for machine-

independent character manipulation was to be

made in the new draft Standard.


c.        Mr Hill had received a letter from Mr Shepherd,

the Computer Journal Algorithms Editor, indica-

ting that he would take up the Group's offer to

check algorithms against the Standard. One

algorithm had been received by the Secretary.


d.        A copy of the Schneck-Angel optimizing compiler

had been received by the Atlas Computer Labora-

tory. It was to be mounted on the 360/195 and

1906A systems.


3.    CHAIRMAN'S REMARKS       A history of the British Computer Society was to be

written by Mr Murray Laver and Mr Harold Gearing was

to help collect information for it. Any contri-

butions, particularly about the early years of the

Society, were welcome and should be addressed to

Mr Gearing at Society Headquarters.

Attention was drawn to the facts that Datafair was

to be held again in 1975 and that John Player Computer

Award nominations were now due.


4.    BSI FORTRAN              Dr Day had received a genuine apology from the Society

ACTIVITIES               (v. minutes of previous meetings) and it was hoped

that the matter had finally been resolved

satisfactorily.


Mr Shearing was now the BCS Fortran representative on

the BSI Programming Languages Committee DPE/13; Dr Day

was also on the committee but did not represent the

BCS.


5.    ECMA FORTRAN             ECMA Fortran Committee TC8 had sent to ANSI proposals

STANDARDS                (1) concerning generic functions, suggesting the

ACTIVITIES               introduction of a new PRIVATE statement and (2) con-

cerning DO loops, suggesting rescinding extended

ranges and real and double precision control variables.

It was anticipated that there would be one more meeting

of TC8 at which decisions affecting the substance of

the draft American Standard could be made.


6.    ANSI FORTRAN             The minutes of the ANSI Fortran Committee, X3J3 meeting

ACTIVITIES               of March 11-14 1974 were considered. The principal

decisions then taken were:


a.    To require list-directed PRINT and WRITE

statements to start a new output record.


b.    To require that a Hollerith constant used in a

DATA statement contain no more characters than

may be held in the corresponding single datum,

and that complex variables may not be

initialized with Hollerith constants.


c.    To specify that in subset level 1 the

characters $ : > < be not provided, that a

maximum of 9 continuation cards be required,

that BLOCK DATA be not provided and that

restrictions be made on the position of DATA

statements within a program unit. Further,

that level 1 should allow all data types

except double precision and complex.


        d.  To specify that subset level 2 should

incorporate the full character set, all data

types and BLOCK DATA subprograms.


It was noted that X3J3 had scheduled meetings

beginning May 20 (Willow Grove PA), July 8 (San

Diego), September 3 (Boulder), November 18 (New

York) and January 13 1975.


7.    ANSI FORTRAN             There was a discussion of the extensions and

OVERVIEW                 revisions made by X3J3 and allocated to subgroups

X3J31, X3J32 and X3J33 (v. Appendix B of minutes of

15.3.74). Comments on a number of proposals were

sent to X3J3 (see Appendix A).


8.    DATE OF NEXT             The next meeting will be held on Friday 28 June 1974

MEETING                  at 10.30 am in the Library, Royal Society of Tropical

Hygiene and Medicine, 26 Portland Place, London W1.

The meeting will end by lunch time. The main item on

the agenda will be a discussion of the X3J3 input-

output proposals.




APPENDIX A



To:  X3J3


From:  David Muxworthy for BCS Fortran Specialist Group


Subject:  Comments on X3J3/50 and FORTREV


1.    Order of Statements (group X3J31)


In accepting proposal 88 (/39,4.8) X3J3 consciously made the following

standard-conforming sequence of code illegal:


DIMENSION A(X)

INTEGER X


According to the proposal voted on, the following standard-conforming

statement would not be allowed, but the wording of FORTREV paragraph 3.5c

does not prohibit it:


INTEGER A(X), X


Further, in the rationale for proposal 88 (/39,G4) it is implied that the

following code does not conform to the 1966 standard:


EQUIVALENCE (D,X(2))

DIMENSION X(10)

DOUBLE PRECISION D


Our interpretation is that this conforms to the present standard, but not

to the proposed new one. Does X3J3 agree that this represents an incom-

patibility with the 1966 standard and what is the intention of the committee

with regard to the second example above?



2.    Hollerith data in complex variables (group X3J33)


Our understanding of the 1966 standard (section 7.2.2) is that complex

variables may be initialized with Hollerith constants so that the decision

(/50,4.2) explicitly not to allow this creates an incompatibility with the

old standard and represents an inconsistency in the new one, since complex

variables may be defined with Hollerith values by a READ statement or by

an assignment statement using the CMPLX function (or, in practice, by being

equivalenced to REAL variables defined with Hollerith values). For

compatibility and consistency we suggest that it should be permissible to

initialize complex variables etc with Hollerith data and, by analogy with

input-output, that in the DATA statement each variable should be associated

with two Hollerith constants.



3.    Control variables in DO-loops (group X3J33)


There was some confusion over whether FORTREV (section 11.6.4(2)) required

the control variable to be incremented by the addition of m3 at each itera-

tion or whether another method, e.g.


control variable = m1 + m3 * (times round - 1)


was allowed by the term "algebraic addition". There was concern over

rounding problems in general in connection with DO-loops and a straw poll

voted 14-6 in favour of allowing only integer control variables.


4.    Length of unformatted records (group X3J34)


The length of an unformatted record is measured in storage units (FORTREV

12.1.1, 12.11.1b2, 12.11.3.3). Unless it is to be prohibited to write

characters to a direct-access unformatted file, the programmer must be

provided with some means of determining the length in storage units of a

record occupying both storage and character units.


5.    Processor-supplied functions (group X3J35)


There was some discussion of the ECMA proposals for combining processor

supplied functions into one list and for a PRIVATE statement (ECMA/TC8/74/9)

In a straw poll there was a 14-3 majority for abolishing the distinction

between intrinsic and basic external functions and a 15-0 majority for

introducing a new specification statement to denote user-supplied functions

rather than extending the EXTERNAL statement.




APPENDIX B



COMMITTEE: Programming Language Committee              DATE: April 25, 1974


TO:        News Release                                ADDRESS WRITER CARE OF:


R.J. Ham

FROM:      R.J. Ham                                    Honeywell Information Systems

     Chairman, PLC                               300 Concord Road

Billerica, Mass. 01821

Mail Station 805A

Phone: (617) 667-3111 X52988

SUBJECT:  CODASYL FORTRAN DBML Committee


NEWS RELEASE


This is to announce the formation of the CODASYL FORTRAN Data Base Manipulation

Language Committee (FORTRAN DBMLC). The FORTRAN DBMLC is responsible for the

development of proposed specifications that will add a data base manipulation

facility to the FORTRAN Language. This facility will be consistent with the

efforts of ANSI X3J3 on FORTRAN and with the efforts of the CODASYL Data

Definition Language Committee on the data base facility.


Dr Chester M. Smith Jr has been named chairman of the committee.


The inaugural meeting is scheduled for:


The COLONNADE Hotel

120 Huntington Avenue

Boston, Massachusetts 02116


Phone: (617) 261-28OO

June 12, 1974

9:00am - 5:00pm


Individual attendees are responsible for their own hotel arrangements, however,

a block of rooms for the night of June 11 has been set aside for those

identifying themselves with the FORTRAN DBMLC meeting. Please indicate your

plans to attend to:      Dr Smith

Pennsylvania State University

Computer Building

University Park, Pennsylvania 16802

Phone: (814) 863-0422


Thank you.


(Sgd) Ronald J. Ham

Chairman, PLC


PS:   In order to obtain the widest distribution of this announcement, you are

invited to circulate this letter to whomever you think may find it

useful.