From Miles.Ellis@etrc.ox.ac.uk Wed Nov 19 19:37:39 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 TAA15622 for ; Wed, 19 Nov 1997 19:37:37 +0100 Received: from ermine.ox.ac.uk by oxmail4 with SMTP (PP) with ESMTP; Wed, 19 Nov 1997 18:37:31 +0000 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 SAA13435 for ; Wed, 19 Nov 1997 18:37:30 GMT Date: Wed, 19 Nov 1997 18:37:30 GMT X-Sender: mellis@ermine.ox.ac.uk Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: "WG5 members (list)" From: Miles Ellis Subject: Interoperability of Fortran with C Dear all, The country vote on the confirmation of Vienna Resolution V4 (to request the primary developent body to take over responsibility for Interoperability with C as a firm requirement for Fortran 2000) ended on Thursday 6th November 1997. The countries who voted were as follows: Finland YES Germany NO Japan YES UK YES USA YES The German NO vote was accompanied by a number of comments which are appended to this message. The Vienna Resolution V4 is, therefore, confirmed and I have informed the Chair of J3 (Jerry Wagener) that WG5 formally requests J3 to take over responsibility for further work in this area as a firm requirement for Fortran 2000. I have already written to Michael Hennecke to thank him for his work up to this point, and I would like to take this opportunity to pay a public tribute to him for all the work that he has put in (with very little constructive help from anyone else, regrettably). My personal view, with the valuable benefit of hindsight, is that this project was probably never a suitable project for a Type 2 Technical Report. At the time when we introduced the concept of these TRs it was very clear that there was an urgent need for the feature (after all we lost our provisional editor precisely because his organisation, CERN, withdrew from Fortran Standardization work largely because of Fortran's inability to utilise libraries written in C!). However, I don't think that any of us realised at the time how much work would be involved and how all-pervading much of the resulting syntax and semantics would be. The requirements for rapid conclusion and orthogonality were, therefore, not met - although we none of us realised that at the time. Despite the difficulties arising from this, Michael made a valiant effort which, I am confident, will greatly help J3 in their task of incorporating the interoperability feature into Fortran 2000 as a high priority requirement. Miles ------------------------------------------------------------- The comments accompanying the German NO vote are as follows: At its meeting last week, the DIN working group for Fortran unanimously agreed that the procedure (or rather the lack thereof) by which WG5 arrived at resolution V4 in Vienna is very questionable. As a consequence, Germany does not regard resolution V4 as a common viewpoint that was discussed and supported by an informed majority. First, the facts about the last WG5 meeting, which are, by now, history: 1. The subgroup for "Interoperability with C", headed by the project editor Michael Hennecke, discusses technical and political issues all week. 2. As a result of these discussions, a resolution is drafted that is supposed to reflect the general opinion of the subgroup and of the working group as a whole. Its wording is established in the course of the week, and the draft remains essentially unchanged until Friday morning. 3. Late Thursday afternoon, the problem of producing the TR on Interoperability with C in a timely manner is discussed, but no decisions about how to continue are made. The new version of the draft resolution produced on Friday morning essentially reverses, in a single sentence, the process for this work - that is, the process WG5 established at the Tokyo meeting concerning the production of three technical reports, including obtaining new work items from SC22 to develop these TRs." 4. Some heads of delegation and members of the drafting committee as well as several other members of WG5 are unaware of the fact that the content of resolution V4 is undergoing drastic changes some time between Thursday evening and Friday morning and are not informed until the revised resolution appears on a new paper version of the resolutions distributed shortly before the final voting on Friday morning. 5. Without any opportunity for more discussion, Germany and all of its delegates are forced to vote against this resolution. Obviously others feel uneasy about this quick and unexpected reversal of established policy, judging from the additional vote against the resolution and two abstentions. The majority of delegates and countries apparently are persuaded by the argument that the risk of failing to produce the TR in time for it to be incorporated into the next revision of the Fortran standard is too great. Germany believes that such a drastic change of policy was by no means necessary and that other solutions could have been found in case WG5 thought that there was a serious time problem. For example, a development/support team could have been formed and a rigorous time plan established to answer all comments and to progress the work. Instead, WG5 chose to dump the whole effort in what can only be described as a panic last-minute decision. Besides the obvious effects of accepting resolution V4, there are a few less obvious negative consequences. In particular, delegating all responsibility to J3 without any further stipulations as to how to proceed or what to do with the existing, confirmed PDTR clearly contradicts previous decisions in this matter. We cannot see any reason why WG5 should not better define the goals and scope that are to be achieved by this work. Furthermore, there is no guarantee whatsoever that the discussions and controversies which will be encountered once the development takes place within J3 will be less complicated to resolve and that work might progress more quickly within J3. On the contrary, there appear to be considerable differences of opinion within J3 as to how best to provide C interoperability in Fortran; this could cause further delays, possibly delaying the whole Fortran 2002 standard. It will thus be completely unacceptable to Germany if WG5 decides to confirm resolution V4. We believe that the whole decision process was seriously flawed, and that such a major decision, essentially killing off an official work item, cannot be made at the last minute without sufficient deliberation and should have been ruled out of order. -------- End of German Comments ------------- Miles Ellis Convenor: ISO/IEC JTC1/SC22/WG5 Email: Miles.Ellis@etrc.ox.ac.uk Phone: +44 1865 270528 Fax: +44 1865 270527