British Computer Society - Fortran Specialist Group
Minutes of Meeting held at BCS Headquarters on Tuesday 28th October 1980
Present: P. A. Clarke (Acting Chairman) Rothamsted Experimental Station
M. R. Dolbear B.P. London
D. H. Marwick Heriot-Watt University
B. Meek Queen Elizabeth College, London
L. P. Meissner ANSI X3J3 (Lawrence Berkeley Labs. USA)
J. D. Murchland 35 Rectory Park, Sanderstead, Surrey
J, A. Murray CCTA
D. Muxworthy Edinburgh University
M. Nunn CCTA
T. L. van Raalte M.O.D.
G. A. Ruscoe Geisco Ltd.
R. Scowen N.P.L.
J. Thomas Geisco Ltd.
D. M. Vallance Salford University
A. Wilson ICL
J. D. Wilson (Secretary) Leicester University
l. Apologies
An apology for absence was received from G. Harding. The chair was taken
by P. Clarke, Vice-Chairman.
2. Minutes of Previous Meeting [8 September 1980]
Item 3 (X3J3 Activities) second paragraph: the item "zero trip DO" and the
following sentence should be deleted. The minutes were then approved.
3. Matters Arising
(i) The full reference for the report mentioned in item 6(ii) is:
"A cross-reference index for the syntax of the programming language
Fortran 77" by R. S. Scowen, NPL Report CSU 5, December 1979.
(ii) A draft report of the NBS study mentioned in Appendix B has now been
produced. This has been submitted to NBS for approval. Copies of the
draft, entitled "Portable Fortran 77 Programming" may be obtained from
J. Larmouth, Salford University.
4. Report from October Meeting of X3J3 - Fort Lauderdale
L. Meissner summarised the main features of the meeting.
A tutorial was held on relations between X3J3 and EWICS (European Workshop
on Industrial Computer Systems - formerly Purdue, Europe).
A proposal to include Array Intrinsics was made by A. Wilson. This would
include array sections, generation functions, shift functions: the proposal was
passed.
A draft proposal on General Precision was made by B. Smith. No syntax has
been formulated as yet. A straw vote was taken but was inconclusive.
A proposal for abbreviation of qualified names (alias) in data structures
was withdrawn pending further discussion.
A tutorial on "extended variant" structures was given.
A proposal to allow Recursive Subroutines and Functions was passed.
A proposal on Name Management via PREFIX was passed.
A proposal to allow blanks in FORMATs was passed.
A proposal to allow list-directed I/O failed.
A draft proposal on Groups and Internal Procedures produced favourable
straw votes.
5. Report from ISO Fortran Experts Meeting - Amsterdam
D. Muxworthy reported from the meeting.
The main presentations were:
Summary of X3J3 Proposals for Fortran 8X - J. Martin & J. Matheny, USA
Comments on X3J3 Proposals - D. Muxworthy, UK
Language Requirements for Numerical Software - J. Reid, UK
Generalised Real Data Type - J. L. Schonfelder, UK
Interface Solution to Application Facility Modules - J. Wagener, USA
Completeness Module - A. Clarke, UK
Standardisation of Industrial Real Time Fortran - W. Kneis, W. Germany
Implementation of Real Time Fortran 8X as Module - J. Snoek, NL
Discussion of Fortran 77 - B. Holberton, USA
Generalisation of Fortran - I. Dahlstrand, Sweden
General or Specific Solution to Multi-Tasking Semantics - K. Ampt, NL
Although still formally committed to a release date of 1985, X3J3 are now
talking in terms of 1988.
The Core plus Modules mechanism was re-stated, although it is not clear
whether the Core will be large or small. The following diagram was produced
by L. Meissner.
The following extensions were discussed:
- global data definition (cf COMMON)
- data structures approved
- Character extensions
- Bit type and BNOT,BAND
- Octal and hexadecimal constants
- environmental enquiry parameters
- free form source, significant blanks
- character set to be extended ASCII
- long names up to 31 characters including underscore
- OPEN and CLOSE extensions
D. Muxworthy referred especially to three presentations:
(i) Interface Solution to Application Modules - J. Wagener, discussed
extended procedure calls, keyword arguments, defaults and macros.
(ii) Completeness Module - A. Clarke. The meeting voted 12 for, 1 against,
11 undecided to continue work on the proposal.
(iii) Discussion of Fortran 77 - B. Holberton, outlined NBS support for
Fortran 77. This includes a Fortran 77 analyser, a flow analysis
program, a suite of test programs and documentation "Help for the
lonely programmer".
6. Any Other Business
D. Muxworthy reported that X3J3 have suggested holding another Fortran
Forum in the UK (preferably London) in October 1981. This would probably be
held on Monday 12th October 1981, after the ISO TC97 SC5 meeting (Oct.5-9, '81)
at ICL Beaumont. The group approved this suggestion in principle and will set
up a committee to organise it. It was pointed out that October 12th will be
during term time so finding a sufficiently large room may prove difficult.
L. Meissner said that Friday October 2nd would be an alternative date but would
be less attractive to X3J3 members due to travel arrangements.
7. Next Meeting
The next meeting is provisionally scheduled for Monday 2nd February 1981
at BCS Headquarters. In the afternoon Steve Hague from the National Algorithms
Group will talk on "Tools for Software Engineering". Confirmation of date and
other details will be sent with the Agenda nearer the date.
---------------------------
Loren Meissner gave a presentation on the US Department of Energy Language
Working Group's report "Fortran Language Requirements". This report was published
in draft form in August 1979 and is currently being revised.
The Language Working Group was set up by the Advanced Computing Committee of
the Dept. of Energy to study and make recommendations concerning language needs
of the Department's Laboratories and Program Offices. It was decided that the
"heavy duty" Computer Laboratories would take the initiative and design extensions
to Fortran 77 (not Fortran 8X) leading to a possible Common Front-End Compiler
implementation. This resulted in the "Red Report". Parts I to IV contain the
function specification; Appendix A contains the language syntax description,
produced by the design team. A special task force was set up to look at static
array features (report UCID - 30175). Many of the members of the Group are also
X3J3 members.
A mnlti-level model is described:
| Nucleus - Fortran 77
|_________________
| Level 1 - Common Features
|______________
| Level 2 - Advanced Features
|________________
| Level 3 - Development Features
|_________________
Level 1 includes:
- word oriented BIT type operations
- compiler directives (turning on & off listing)
- time and date enquiries
- NAMELIST I/O
- asynchronous I/O (start, read, check, continue etc.)
Level 2 includes:
- array processing
- data structures
- macro processing
- dynamic storage
- control structures
- COMPLEX DOUBLE PRECISION
- environmental enquiry
- free source form
- more NAMELIST
Level 3 includes:
- more array processing (array assignment, array section objects,
matrix operations elementwise, means of identifying diagonals
etc.)
There are no plans to go further with this study, beyond revising the
Red Report.
J. D. Wilson
November 1980