Project 3: Critique of Fortran Standards This project "Critique of Fortran Standards" was soon dropped in view of other interactions with the Fortan community. See project 20. -------------------------------------------------------------------- Oak Brook 1976: =============== Dr. Reid reported that there had been a disappointing response to the document he had sent to the FORTRAN Standards committee. There was strong feeling in the Working Group that some parts of the proposals should be resubmitted, especially that for double precision complex since, as pointed out by Dr Lawson, the committee was now open to comment upon their present draft with a deadline of July 1976. It was also agreed that any further proposals should be sent in as representing the views of WG 2.5 even though they could not form an official IFIP document. It was noted that Ms. Holberton (a member of the X3J3 FORTRAN Standards Committee) would be present at the Portability Workshop. The group decided to review the position at a later session in the light of any comments which might be made at the Workshop. ---- Dr. Reinsch reported the discussions following the Portability Workshop on minimal hardware requirements. Mr. Du Croz had acted as secretary during the working session and would be writing and circulating minutes. The session had been too short to reach any decisions, but a number of aspects had been briefly examined: 1. Hardware 1.1 Number Systems 1.2 Basic Operations 1.3 Arithmetic Relations 1.4 Overflow/Underflow/Feasible Operations 1.5 Integer Arithmetic 1.6 Documentation Topics yet to be discussed included: 2. Software 2.1 Elementary Operations 2.2 Extended Arithmetic Facilities 2.3 Interface to Compilers 2.4 Precision Control The work of the section would continue by written communication. Professor Hull felt that the report had failed to include the enthusiasm of the discussions. A grading system for existing machines had been evolved, and decisions made regarding the base of representation, the method of rounding and the handling of unusual numbers, underflow and overflow. Dr. Reid asked what the timescales were for activity. Dr. Reinsch hoped to have an intermediate document ready for the next meeting. Professor Kahan in particular was keen that the ultimate document was more than a list; there was a need for delineation of the practical reasons for all the decisions reached, and for this supporting evidence to be properly documented. Professor Stetter suggested that papers for discussion be made available at least two months prior to the next meeting. ---- There was detailed discussion of the latest draft of the proposal on INTRINSIC functions, for submission to X3J3, It was further agreed that the proposals on COMPLEX DOUBLE PRECISION and on MAP should be submitted to the FORTRAN Standard Committee. The Secretary agreed to draft suitable covering letters. The letter covering the document on Intrinsic Functions would include the names and affiliations of the members of the Working Group present, together with those of the original drafting committee and assisting secretariat. Dr. Lawson proposed that the Working Group designate its own contact for with X313. It was unanimously agreed that the Chairman of WG 2.5 should serve this function. Amsterdam 1977: =============== Dr Reid noted the debt of the Group to Professor Moler for his work, with others, in preparing the document on intrinsic functions. The note proposing the addition of Double Precision Complex to the variable types in the draft proposed FORTRAN report owed its detail to the encouragement and advice of Ms Betty Holberton, a member of the X313 committee. The paper on the MAP statement was the work of Doctors Ford and Smith (particularly Smith). Unfortunately all the proposals were rejected. The social factors in the X313 committee that made this always the likely result were discussed by Dr Feldman, in a letter to Doctors Ford, Reid and Smith (see Appendix 3.2, particularly the lined paragraph). Dr Meissner in correspondence to the same members noted that the proposals were carefully prepared (see Appendix 3.3, lined paragraph 1) and gave a list of reasons that may have caused their rejection (see Appendix 3.3, lined paragraph 2). He went on to propose how the Group might best attempt to affect the future development of FORTRAN (see Appendix 3.3, lined paragraph 3). In discussion it was agreed that there was a basic need for a representative of the numerical software community to joint X3J3. Dr Brown commented on the heavy workload and time commitment (a week meeting each alternate month) membership of the committee brought. Perhaps inevitably this caused the committee to be dominated by representatives of the computer manufacturers. Dr Reid asked what further action should be taken with the proposals. It was agreed to continue to press for the inclusion of the double precision complex data type in FORTREV 77. Members of WG 2.5 were asked to encourage the implementation of the facilities of the intrinsic functions in compilers for all languages. MAP would be left in the public arena for discussion and debate. It was unanimously agreed (proposed Dr Reid, seconded Professor Hull) that 'the Working Group would appreciate and encourage the funding of a person representing the interests of the Numerical Software community (perhaps even a member of WG 2.5) joining the X3J3 Committee'. The Chairman thanked Doctors Smith, Reid and Ford, amongst others, for their work in preparing and submitting papers from WG 2.5 to the X313 committee. --- Dr Reid proposed a program of work on future Fortran (i) Double Precision Complex He would prepare a new document with Dr Smith, which would first be sent to Dr Brown. An agreed document would then be sent to all Working Group members for their comments. (ii) Intrinsic Functions A document on intrinsic functions would be drafted by Dr Brown This would not address the question of environmental parameters which had been dealt with separately. The paper would be in a language independent form. The document would be sent Dr Reid and Dr Smith in the first instance, and when agreed dispatched to all Working Group members for their comments. (iii) MAP This proposal would be left untouched for the present. The document was in the public arena and would be left to ferment. Dr Reid felt that the Group should be looking at how the proposals emerging from the discussion of preferred hardware features would affect and should be implemented in a future FORTRAN. The three members of the Group involved in preparing the draft documents undertook to prepare and agreed the papers as quickly as possible. After comment and amendment by Group members the documents would be offered for publication in SIGNUM Newsletter, in the name of the authors, as a working paper of WG 2.5. The program of work and mode of operation was agreed by the Group.