BRITISH COMPUTER SOCIETY - FORTRAN SPECIALIST GROUP


MINUTES of a Meeting held on Monday 2nd October 1978 at Imperial College


PRESENT:


P A Clarke (Chairman)                Rothamsted Experimental Station

G L Harding                               E C M W F

W S Hilder                                Computer Systems Portability

D Hill                                        Central Computer agency

D J Holmes                               Rolls Royce Ltd., Bristol

J D Murchland                           University College London

T L van Raalte                           Ministry of Defence

A Wilson                                   International Computers Ltd

J D Wilson                                University of Leicester

J M Roberts-Jones (Secretary)    Liverpool City Council


APOLOGIES FOR ABSENCE:


D T Muxworthy                          Edinburgh R C C


1        The minutes of the previous meeting were approved. [21 August 1978]


2        There was a discussion of the possible reasons for low attendance

at meetings. One suggestion was that more publicity was needed.

The Chairman reported that he had recently had fruitful

discussions with Malcolm Peltu, the Editor of Computer Weekly,

who was sympathetic to the cause of standards activities and was

likely to assist us whenever possible. It was questioned whether

most programmers were aware of Fortran 77; many possibly knew

only of the facilities provides by their current compilers. Some

doubts were expressed as to whether greater awareness of standards

would necessarily be beneficial.


3        It was reported that copies of the new standard could be ordered

from the British Standards Institute, Overseas Sales Branch,

101 Pentonville Road, London N1 9ND, at £17.80 per copy including

postage. Payment is required with the order. The title should

be quoted in full, as "American National Standard X3.9-1978

Programming Language Fortran". It was hoped to negotiate a

discount for a bulk purchase by the Group.


4        It was noted with some regret that the first Fortran compilers to

appear all included various extensions. It was felt that there

was an urgent need for a Fortran 77 verifier analogous to PFORT.


5        It was noted that it was proposed to form a Technical Committee

on Fortran under ACM Sigplan, the Chairman to be Jerry Wagener.


6        The Chairman led a discussion on recent changes to

Codasyl database specifications, with particular reference to the

question of defaults for omitted keyword parameters in the

Fortran DML. It was suggested that there seemed to be a danger

of the FDBLC introducing too many new conventions too soon.


7        The proposals to be submitted to the ISO meeting by BSI/DPS13/WG6

generated much debate as to the proper aims of the next revision.

In particular, two distinct schools of thought emerged on the

respective merits of the probably incompatible objectives of a

language "that does what users really want" and of a language

tuned for maximum portability. It was suggested that a lack of

upward compatibility would be acceptable if and only if portable

Fortran 77 programs were made available for source conversion.


8        The Steering Committee discussed plans for a meeting in December,

at which it is hoped that several members of X3J3 and maybe some

representatives from other countries will speak on Fortran 77 and

Fortran 82. A charge will be made for attendance, catering will

be arranged, and a copy of the new standard will be supplied to

each delegate.


[A Fortran Forum was held in London on 1 December 1978, following the
initial ISO Fortran Experts' Group meeting. The next Fortran Specialist
Group meeting
was held on 5 February 1979.]




APPENDIX A


FORTRAN DBLC - MINUTES OF AUGUST 1978 MEETING.


It was noted that eight features, absent from Fortran 77, would

be of significant assistance in developing a Fortran database interface

specification:


*        Data structures

*        Error handling

*        Intrinsic subroutines

*        An INVOKE statement

*        I/O list parameters

*        Keyword parameters

*        Longer names


*        Varying arrays


In anticipation of the formation of an X3J3 Technical Subcommittee

comprising three X3J3 members (Greenfield, Signor, Upshaw) and two FDBLC

members (Drake (chair), Greenwood), it was decided to hold future meetings

at the same location as X3J3, with a four-day overlap.


A List was compiled, of problems required to be resolved before

publication of version 2.0 of the Journal of Development. Many of the

items are caused by the major revisions earlier in 1978 to the

specifications of the Schema and of the Cobol Interface. Investigation

was needed into the consequences of various changes to the Cobol DML

and subschema:


*        Realms need not be disjoint

*        The verbs ERASE, CONNECT and DISCONNECT had been redefined

*        The effect of the MODIFY and READY verbs on currency indicators

          had been changed

*        There might be inconsistencies as between the Cobol and Fortran

          specifications of record selection expressions

*        Revised rules for updating currency indicators (with a feeling

          that the new Cobol rules, while being an improvement, are

          still too complex and require further changes)

*        Revised record locking specifications and introduction of

          commitment units (COMMIT, ROLLBACK, KEEP and FREE verbs)


It would also be necessary to investigate the inconsistencies as

between the Cobol and Fortran rules for updating special registers, to

remove references to record storage location modes and CALC keys to

reflect the removal of specification of physical attributes from the

Schema DDL; to recast the ACCEPT functionality as a Fortran function,

and to investigate possible changes in two areas where the present

specification is obviously inconsistent with "the flavour of Fortran",

i.e. special registers and CALL BEFORE/AFTER phrases.




APPENDIX B


CODASYL DATABASE FACILITY.


The current published Codasyl specifications comprise one

document defining the Schema DBL, and two defining the Cobol and

Fortran interfaces respectively.


"Codasyl Data Description Language Committee Journal of

Development 1978". Available from

Materiel Data Management Branch,

Department of Supply and Services,

4th Floor, Core B1,

Place du Portage, Phase III,

11 Laurier Street,

Hull, Quebec,

Canada


Specify title exactly as quoted; remittance of $6.59 in U S or Canadian

currency with order.


This document replaces the 1973 JOD, and specifies the Schema DDL.

Major changes include:


*        Database keys are redefined as unique record references only

 within a single rununit

*        The DDL is restricted to describing the logical structure of a

 database: facilities for describing the physical realisation

 have been deleted (SEARCH KEY, LOCATION MODE, ACTUAL/VIRTUAL,

 PRIOR PROCESSABLE, LINKED TO OWNER, index declaration)

*        Recursive sets are now allowed

*        A new RECORD KEY clause allows declaration of identifiers as

 record order keys independently of set relationships

*        Value-based set relationships may be defined by STRUCTURAL

 CONSTRAINT clauses

*        ENCODE/DECODE, dynamic sets, temporary areas are deleted

*        A record may be specified as WITHIN ANY AREA


An appendix contains a draft specification of a Data Storage

Description Language; with facilities for describing a Storage Schema

defining the physical representation of entities in a schema.


The Cobol interface is integrated into the Codasyl Cobol JOD 1978,

available to BCS members at £7 (cash with order) from BCS HQ.


The Fortran interface is described in "Codasyl Fortran Data Base

Facility 110-GP-2 January 1977", available from Materiel Data Management

Centre, Department of Supply and Services, Ottawa, Ontario, Canada.

Specify title exactly as quoted; remittance of $4.00 in US or Canadian

currency with order.