From Miles.Ellis@etrc.ox.ac.uk Tue Oct 7 17:45:01 1997 Received: from oxmail4.ox.ac.uk (oxmail4.ox.ac.uk [163.1.2.33]) by dkuug.dk (8.6.12/8.6.12) with SMTP id RAA14576 for ; Tue, 7 Oct 1997 17:44:59 +0100 Received: from ermine.ox.ac.uk by oxmail4 with SMTP (PP) with ESMTP; Tue, 7 Oct 1997 17:44:07 +0100 Received: from [163.1.85.1] (com1.etrc.ox.ac.uk [163.1.85.1]) by ermine.ox.ac.uk (1.1/8.8.3) with ESMTP id RAA27214 for ; Tue, 7 Oct 1997 17:44:02 +0100 (BST) X-Sender: mellis@ermine.ox.ac.uk Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Tue, 7 Oct 1997 17:44:57 +0100 To: "WG5 members (list)" From: Miles Ellis Subject: Country Ballot on PDTR 15815 - Interoperability with C Further to my email of about two weeks ago, I have finally put the results of the SC22 ballot on the WG5 server as document N1305. It is in pdf, html, PostScript and text forms - all based on the email distribution from the SC22 Secretariat. It takes up 34 A4 pages! I am sorry for the delay. I have been out of the office quite a lot during the last two weeks and I had thought that I had already put the ballot results on the server; I only realised today that I had forgotten! However, we still have to decide what to do next, and I have discussed this with several people, including the SC22 Secretariat. The consensus of those to whom I have spoken is that the WG5 Resolution clearly takes precedence over the SC22 ballot, since most countries voting in the SC22 ballot would not be aware of the decision taken by WG5, and even if they were it might not (even should not ;-) have altered their view of the document that was before them. Nevertheless, there were several active members of WG5 who were not in Vienna, and as I said at the time, I wish to give them a chance to provide their input. The situation is that (a) in Vienna, WG5 resolved (V4) That WG5, anticipating considerable technical revision of the proposal on Interoperability of Fortran and C as a result of comments in the SC22 ballot on PDTR 15815, and in order to keep to the schedule for implementing interoperability as a major requirement for Fortran 2000, requests the primary development body to accept the responsibility for developing this requirement. (b) SC22 has voted to register the document by 14 votes to 2, with 2 abstentions, and to approve it as a PDTR by the same margin. In both cases the NO votes came from the UK and USA, while France and Germany accompanied their YES votes for approval with a number of comments. The complete set of comments is contained in WG5 N1305. We now have to either confirm our decision in Vienna, or reverse it and continue as before. The advantages and disadvantages of each approach, as I see them, are as follows: A. CONFIRM our Vienna decision and pass the responsibility for incorporating interoperability with C to J3 as a major Fortran 2000 requirement. The advantage of doing this is that J3 have indicated that they are willing to accept this responsibility, and have already allocated resources to the work. If we adopt this approach then Fortran 2000 WILL have some form of interoperability with C. The only disadvantage that I can see to this approach is that I shall have to ask SC22 to cancel the project. This is no big deal - other WGs cancel projects from time to time, sometimes at a much later stage than this. (Last year one WG requested permission to cancel a project after the document had already passed its DIS ballot! The reason was that the WG concerned had come to the conclusion that the standard was no longer appropriate, notwithstanding the approval at the SC22 and JTC1 levels. Their request was approved.) B. REVERSE our Vienna decision and attempt to complete the processing of the TR as originally planned, with Michael Hennecke as Project Editor. The only advantage that I can see in doing this now is that we do not have to ask SC22 to allow us to cancel the project. There are several disadvantages: (1) In Vienna Michael indicated that he would need a substantial amount of help in processing the comments that were anticipated - although I don't think that anyone had expected 34 pages of comments. I am not aware of a rush of people offering to help him, and without such assistance the task is almost certainly too much to expect Michael to deal with on his own. [Note that J3 members already have a substantial workload associated with the development of Fortran 2000. It is one thing to allocate some of their J3 time to Interoperability, and quite another to take on Interoperability work in addition to their J3 workload.] (2) The TR process was introduced in order to allow a small number of key issues to be developed ahead of the next revision so that they could be used for some time before incorporation into the standard itself during that revision. Given the lack of resources it seems most unlikely that the TR could be submitted for a DTR ballot before late next year, and it would be a brave man (or woman ;-) who would expect that ballot not to attract any negative comments and votes. It would seem highly unlikely, therefore, that any TR could be published until the second half of 1999 at the earliest. This will be too late for it to be incorporated into Fortran 2000 since it is unlikely that there will be any compilers implementing the proposals for some time after the publication of the TR. The most optimistic result, therefore, would be that the TR would co-exist with Fortran 2000 - although whether any vendors would wish to implement it at the same time as upgrading their compilers to Fortran 2000 specification is doubtful - and then be incorporated into the revision of Fortran 2000. The worst case scenario is that the TR would be simply ignored, as vendors either include their own, non-standard approach to interoperability, if their customers demand it, or simply ignore the whole issue. (3) Although it was a difficult decision, and one which caused many people some angst, we did decide in Vienna that the only way forward was to roll up this project as part of Fortran 2000. Nothing has been said by anyone since then to suggest that things have suddenly got easier. I very much suspect that if we reverse our Vienna decision now then we shall be faced with the same problem a bit further down the line. Only then it will be too late to stand any chance of getting interoperability into Fortran 2000. Personally, I believe that if we wish to have a feature in Fortran 2000 allowing interoperability with C there is now only one way to achieve that - by confirming our Vienna decision and accepting J3's offer (made at our request) to take over development of the feature from Michael Hennecke as a firm requirement for Fortran 2000. I am very much afraid that if we do not do this then the project will be delayed to the extent that the earliest that it will be available is some time AFTER Fortran 2000, and maybe never. I wish that it was not so, especially after all the work that Michael has already put into the developent of the current document, but I think that ensuring the incorporation of the feature into the base Fortran language is the most important thing at this time. I emphasise that the above are my personal views. Obviously, I shall do whatever the Working Group as a whole wishes. In order to determine what is the wish of WG5 in this matter, therefore, I plan to carry out a COUNTRY ballot to confirm, or otherwise, our Vienna decision on this topic. J3 have their next meeting in the week 10-14 November, and I shall therefore close the WG5 ballot 30 days from today, on Thursday 6th November, which will (just) allow me to communicate the result to J3 in time for them to take note (and any necessary action) at their meeting. Miles -------------------------------------------------------------------------- THIS IS A COUNTRY BALLOT - ONLY HEADS OF NATIONAL DELEGATIONS MAY VOTE ______ Do you wish to confirm the position taken in Vienna | | Resolution V4 to request the primary developent body | YES | to take over responsibility for Interoperability with |______| C as a firm requirement for Fortran 2000? | | [The primary development body has already agreed to | NO | do so, and has allocated resources for this work.] |______| (signed) ....................................... on behalf of (country) ......................... PLEASE RETURN THIS BALLOT TO THE CONVENOR AT NO LATER THAN 1700 GMT ON THURSDAY 6TH NOVEMBER 1997 --------------------------------------------------------------------------- ------------------------------------------------------------ Dr Miles Ellis Director: Educational Technology Resources Centre University of Oxford, 37 Wellington Square, Oxford OX1 2JF, ENGLAND Telephone: +44 1865 270528 Fax: +44 1865 270527 Email: Miles.Ellis@etrc.ox.ac.uk WWW: http://www.etrc.ox.ac.uk/Personal/Miles/Miles_Ellis.html