From mellis@vax.ox.ac.uk Mon Jul 19 13:02:12 1993 Received: from oxmail.ox.ac.uk ([129.67.1.1]) by dkuug.dk with SMTP id AA02912 (5.65c8/IDA-1.4.4j for ); Mon, 19 Jul 1993 13:00:50 +0200 Received: from vax.ox.ac.uk by oxmail.ox.ac.uk with SMTP (PP) id <02431-0@oxmail.ox.ac.uk>; Mon, 19 Jul 1993 12:01:02 +0100 Received: by vax.ox.ac.uk (MX V3.3 VAX) id 12804; Mon, 19 Jul 1993 12:02:16 +0100 Date: Mon, 19 Jul 1993 12:02:12 +0100 From: Miles Ellis To: SC22WG5@dkuug.dk Cc: mellis@vax.ox.ac.uk Message-Id: <0096FBB5.00E1D15E.12804@vax.ox.ac.uk> Subject: X3J3 Action on WG5 Resolution B9 X-Charset: ASCII X-Char-Esc: 29 At last week's meeting of X3J3 the resolutions from the Berchtesgarten WG5 meeting were discussed and appropriate action agreed. In particular, the content of resolution B9 (Content of the 1995 Revision) was, of course, examined in considerable detail. After this examination, which involved both subgroup and full committee discussions, the following statement was unanimously approved: X3J3/93-231r1 X3J3 Consideration of WG5 B9 item C7 In accordance with the WG5 request for more information from X3J3 on the "C" list of items in resolution B9, X3J3 offers this early recommendation on the lowest priority item, C7 - derived type I/O. If there are additional ideas from other members of the Fortran community for this item for the 1995 revision, X3J3 would be happy to reconsider this recommendation. C7 is a difficult problem, whose solution appears at this point to require significant resources, and, therefore, X3J3 recommends that this item be ================================= deferred to the F2000 revision. X3J3 also notes that this action is ============================== consistent with the straw votes on this topic at the 1993 WG5 meeting in which no-one supported adding derived type I/O in the 1995 revision, while 25 individuals supported its inclusion in the F2000 revision. When the derived type facility was first developed and adopted for Fortran 90, the issue of input and output of derived type values was given serious consideration. Because no acceptable solution was found at that time, Fortran 90 does not include derived type I/O. X3J3 knows of no subsequent progress on this issue, and initial consideration of the topic at meeting 126 suggests that its resolution is likely to require considerable resources. Further, X3J3 believes that derived type I/O is sufficiently orthogonal to the other aspects of data abstraction and object-oriented programming that deferring work on C7 will not adversely impact the development of a coherent overall solution in these areas. X3J3 concurs with WG5's view, as expressed in the straw votes on individual topics, that this problem should be solved in the F2000 revision, and will plan accordingly in discharging its responsibilities as the primary development body for international Fortran standardization. This statement wil be distributed by the WG5 Convenor in the next distribution, but X3J3 asked me to send this to the WG5 email distribution list as soon as possible after the meeting in order that WG5 members were aware of X3J3's decision on this topic. As a PERSONAL comment, I would add that, with hindsight it is not entirely clear how this item came to be in the content of Resolution B9 at all, since NO-ONE seems to have asked for it to be in the 1995 revision! Nevertheless, we did discuss it, together with all the other items in B9, in an attempt to determine how we would approach it and what resources could be allocated to the issue. The result was the above policy statement. Miles Ellis Director: Educational Technology Resources Centre University of Oxford 37-41 Wellington Square, Oxford OX1 2JF Telephone: +44 865 270528 Fax: +44 865 270527 Email: MEllis@vax.ox.ac.uk