ISO/IEC JTC/SC22/WG5 N1810 Minutes of Meeting of ISO/IEC JTC1/SC22/WG5 Hosted by INCITS/PL22.3, the US Member Body for JTC1/SC22 in Las Vegas, NV, USA February 15-19, 2010 List of Participants: John Reid (JKR Associates, UK) convenor Bill Long (Cray, USA) PL22.3 acting chair Reinhold Bader (Leibniz Supercomputing Centre, Germany) Malcolm Cohen (NAG, UK) Bob Corbett (Oracle, USA) Aleks Donev (Lawrence Berkeley National Laboratory, USA) Nick Maclaren (University of Cambridge, UK) Jeanne Martin (formerly Lawrence Livermore National Laboratory, USA) Toon Moene (Gnu Fortran, Netherlands) David Muxworthy (BSI, UK) Craig Rasmussen (Los Alamos National Laboratory, USA) Van Snyder (Caltech/JPL, USA) Masayuki Takata (Edogawa University, Japan) Stan Whitlock (Intel, USA) Jim Xia (IBM, Canada) Note: WG5 papers are referenced as Nnnnn. They are available from http://www.nag.co.uk/SC22WG5/ INCITS/PL22.3 is abbreviated throughout to PL22.3 and its papers are referenced as 09-nnn or 10-nnn. They are available from http://www.j3-fortran.org/ 1. Opening of the Meeting The meeting opened at 08:00 on Monday 15th May 2010. 2. Opening business 2.1 Introductory remarks from the Convenor Apologies had been received from Dan Nagle who had been prevented from travelling to the meeting for health reasons. Best wishes were sent for his speedy recovery. The convenor said that the main objective of the meeting was to develop the current draft document so that it would be completed for forwarding for an FDIS ballot shortly after the meeting. Other priorities were to review the latest set of interpretations for Fortran 2003 and to review the second draft of the Interoperability TR. If time permitted the Fortran annex of the Vulnerabilities TR would also be considered. 2.2 Welcome from the Host Van Snyder welcomed participants on behalf of PL22.3. 2.3 Local arrangements On behalf of PL22.3 Van Snyder graciously invited participants to dinner at Battista's on the Wednesday evening. 2.4 Appointments for this meeting The drafting committee would be Reinhold Bader, Toon Moene, David Muxworthy (chair), Van Snyder, Masayuki Takata and Jim Xia. David Muxworthy would act as secretary and John Reid as librarian. 2.5 Adoption of the Agenda (N1794) The preliminary agenda was adopted. 2.6 Approval of the Minutes of the Las Vegas 2009 Meeting (N1779) The minutes of the previous meeting were approved. 3. Matters arising from the minutes There were no items not otherwise on the agenda. 4. Status of Las Vegas 2009 Resolutions (N1778) It was noted (cf. LV3) that Technical Corrigendum 4 had been published and that the minor error in the document had not delayed its processing. Otherwise there were no outstanding issues. 5. Reports 5.1 SC22 Matters (Convenor) The convenor drew attention to points in his report of the 2009 SC22 meeting (N1792). A Technical Corrigendum 5 for Fortran 2003 would be produced as a WG5 document but would not be forwarded to SC22. Edits from any outstanding interpretations for Fortran 2003 thereafter would not be collected into a corrigendum. An extension for the schedule for the Interoperability TR had been granted but the timetable would need further review. There was a short discussion on the templates used for member body comments in SC22 ballots. It was recommended that the Excel-based template (N1797) be used. [Secretary's note: The Word-based template must still be used for JTC1 as distinct from SC22 ballots.] There was also a short discussion on paper N1793 on the Netherlands' concerns about the growth of programming language standards. 5.2 National Activity Reports (Heads of Delegations) Canada: The SCC Languages Committee last held a meeting in December. Germany: There were currently only two active members. Japan: Report is in N1809. Netherlands: There had been a teleconference of five members to prepare the Netherlands vote on the FCD; Toon Moene had not been able to attend. A written report (N1819) was submitted later by email. UK: Report is in N1807. US: The US TAG met only at PL22.3 meetings; there were eight members. 5.3 Report from Primary Development Body (INCITS/PL22.3) Development had been proceeding to schedule. 5.4 Reports from other Development Bodies (Editors/Heads) Interoperability TR: Bill Long A second draft of the TR had been produced for this meeting. 5.5 Liaison Reports: INCITS/PL22.11 (C): Craig Rasmussen Discussions with WG14 about the Interoperability TR were continuing so as to ensure that the TR and the C standard were in as close agreement as possible. MPI: Craig Rasmussen The MPI Forum was working to get its new standard more closely aligned with Fortran. Further information was available from Craig Rasmussen. UPC: Dan Nagle No information was to hand. IFIP/WG2.5: Van Snyder There had been a meeting of WG2.5 in Raleigh in August. It had been reported that 600 students at North Carolina University were being taught Fortran 2003. OpenMp: Matthijs van Waveren Report N1817 was submitted by email. Nick Maclaren was giving a course on OpenMP and noted that much of it was written in terms of Fortran 77 rather than Fortran 90. WG23 (Vulnerabilities): Dan Nagle DTR 24772 'Guidance to Avoiding Vulnerabilities in Programming Languages through Language Selection and Use' (N1799) was currently out for SC22 ballot, with deadline March 23. [PL22.3 Opening business from 08:45 to 09:00] 6. Consider the Final CD Ballot comments on Fortran 2008, decide on changes, and construct a response document. There was a preliminary review of the ballot responses in N1802 and the papers referenced therefrom. - GB-1 (proposed work item on portability) Discussion was deferred to Wednesday. - GB-2 (coarrays) David Muxworthy said that this vote had been used to record the strongly held view of the majority of the UK Fortran panel but it was recognized that this opinion had been outvoted in previous WG5 meetings and that there was no need for further discussion unless another member body wished to raise the matter. - GB-3 (delete arithmetic IF) There was opposition to this proposal which would make some existing standard conforming programs non-standard conforming. It was suggested that instead the standard be corrected to eliminate the uncertainty when the value of the expression was a NaN. A preliminary straw vote was: delete the statement 3 - correct it 5 - undecided 6. After further discussion a repeat vote was: delete the statement 3 - correct it 11 - undecided 0. - GB-4 (rename ALL STOP) The proposal was to rename to ABORT ALL. The meeting discussed alternatives, settling on ERROR STOP as preferred alternative. A straw vote then gave: retain ALL STOP 1 - replace by ERROR STOP 12 - undecided 1. There was a further discussion on whether the space between ERROR and STOP be optional. A straw vote gave: require space 11 - space optional 0 - undecided 3. - US-6 (scoping unit fixes) It was noted that although this was categorized as 'technical' in N1802, it proposed changes only to descriptions and made no technical changes. - GB-14. It was remarked that this proposal was incomplete as it specified processor dependencies which should have been added to Annex A. All other comments on the FCD were referred to PL22.3 for detailed consideration. [PL22.3 plenary and subgroups from 10:15 to 17:00] ........................................................................... Tuesday [PL22.3 plenary from 08:00 to 09:55] John Reid drew attention to the following new document on the server: N1809 Japan's National Activity Report and to drafts of the following documents: N1808 Draft TR on "Further Interoperability of Fortran with C" N1813 Responses to FCD comments N1816 Result of WG5 letter ballot on Fortran 2003 interpretations N1817 OpenMP ARB Liaison Report N1818 Reply to comments on interop TR ballot (basically a copy of 10-122) He particularly asked members to check N1813. The meeting then divided into three subgroups: two to consider the responses in document N1816 to the interpretation ballot in N1805 and one to consider the second draft of the Interoperability TR N1808. [From this point on, agenda items 6 (FCD ballot response), 7 (Interoperability TR), 8 (Interpretations) and 10 (Strategic Plan) were progressed in parallel.] The meeting reconvened in PL22.3 mode from 16:30 to 16:40. John Reid drew attention to new drafts of documents: N1812 Strategic Plans N1816 Result of WG5 letter ballot on Fortran 2003 interpretations for discussion the following day: The meeting recessed at 16:45 ........................................................................... Wednesday [PL22.3 plenary from 08:00 to 08:40] There was a short discussion on the draft of the Strategic Plans document (N1812); various minor edits were suggested. It was thought that discussion of revision of Fortran 2008 would be an appropriate topic for the June 2011 meeting. The section relating to the TR on Further Interoperability with C would not be considered until later in the meeting. There was a discussion Nick Maclaren's paper 'Objectivesr1' which had not yet been allocated a number [after revision it became 10-142 and then N1820]. This sought to define clearer objectives for the Further Interoperability with C project. It was apparent also that the operational procedures for the project were unclear. Therefore it was decided that the TR development body should be PL22.3 itself, and not as defined in London resolution L3, and that the mode of operation be similar to that for the base language standard, viz. that WG5 set the requirements and that PL22.3 implement them. The next task therefore was to develop this paper into a requirement document for consideration by WG5. It was reported during the discussion that the second draft TR (N1808) was a great improvement on the first. There followed a tutorial, with discussion, on the proposals in the draft TR by Bill Long. The meeting recessed to subgroup mode (interoperability and interpretations) at 11:35 and reconvened at 16:30. [PL22.3 plenary 16:30 to 16:35] John Reid asked that the revised drafts of N1811, N1812, N1813 and N1816 be reviewed for discussion the following day. ........................................................................... Thursday [PL22.3 mode from 08:00 to 08:10] David Muxworthy introduced a discussion on ballot comment GB-01 in N1802 which called for a new project to define a portable subset of Fortran 2008. In the view of the UK panel Fortran 2008 was failing in its main objective, "to promote portability, reliability, maintainability, and efficient execution of Fortran programs for use on a variety of computing systems". It became clear during the discussion that defining the content of such a subset would be difficult since vendors' implementations were advancing continually and that some of the perceived problems would reduce after a time. Also, there were differing views on whether such a subset should be objective or instructional and on whether such issues would be better addressed in text books or papers than in a standard or a TR. Attention was drawn to document N1793, which had raised related issues at the 2009 SC22 meeting. It was suggested, facetiously but pertinently, that promoting portability be dropped from subclause 1.1 of the standard. A straw vote on whether there should be a new project to define a subset went: yes 3 - no 9 - undecided 3. 11.1 Future meetings (taken out of agenda order) It was decided that the next meeting would be on June 27 to July 1, 2011 at Garching, Germany, hosted by Reinhold Bader and LRZ. The following meeting would be in June 2012, with exact dates to be confirmed, at Markham, Ontario, Canada, hosted by Jim Xia and IBM. 7. Review the draft TR on further interoperability of Fortran with C and decide on changes. There was a long technical discussion on paper 10-142 "C Interoperability Objectives". Guidance was given to the interoperability subgroup by a series of straw votes: - Should INTENT(OUT) allocatables be allowed? yes 11 - no 1 - undecided 3. Constraint 5 was therefore to be removed from the document. - Should a mechanism be added to get an array element address from subscript values? yes 11 - no 0 - undecided 4. - Should assumed rank dummy arguments be usable for further purposes in Fortran? yes 4 - no 6 - undecided 5. Therefore the TR should not address these things but should leave open the possibility for future development. - Should single descriptors be used rather than C and Fortran descriptors? yes 11 - no 0 - undecided 4. - Should the TR be extended to allow interoperability of assumed length characters? yes 6 - no 4 - undecided 5 by individuals and 2 - 1- 3 by member bodies. Again the conclusion was that this be left for possible future development. - Should the final array in a C descriptor be changed to be of flexible size? yes 10 - no 0 - undecided 5. - Should names for supported types and/or ranks be investigated? yes 11 - no 0 - undecided 4. The interoperability subgroup was to meet later in the day to revise the paper in the light of the discussion. The group also revised the draft TR (N1808) during the meeting. There was further discussion of the draft of the Strategic Plan, N1812, so far as it related to the Further Interoperability TR. It was decided to extend the timescale so that the PDTR would be forwarded to SC22 for ballot in December 2010 rather than April 2010. The meeting recessed to subgroup mode (interoperability and interpretations) at 11:50 and reconvened at 16:30. [PL22.3 plenary 16:30 to 16:45] There was a short discussion on the status of the resolutions and the draft technical corrigendum which was being developed from the interpretations that had been approved during the meeting. [PL22.3 plenary 16:50 to 17:00] ........................................................................... Friday [PL22.3 closing business 08:00 to 08:05] It was agreed that the draft Technical Corrigendum 5 would be checked by a subgroup of Malcolm Cohen, David Muxworthy, John Reid and Stan Whitlock and then made available on the WG5 server. It would not be subject to a WG5 letter ballot as had sometimes happened with previous versions. The drafts of the resolutions (N1811), the Strategic Plan (N1812) and the Response to Comments on the FCD document (N1813) were all discussed and minor changes made. There was a discussion on papers 10-122r1 (Reply to comments on interop TR ballot) and 10-142r1 (C Interoperability Objectives). After revision in the light of the discussion these would become WG5 papers N1818 and N1820 respectively. 9. Construct a draft Fortran annex for the TR on "Guidance to Avoiding Vulnerabilities in Programming Languages through Language Selection and Use". In the absence of Dan Nagle, a member of WG23, no-one present was able to pursue this item. 10.1 Goals for 2010-2013 Van Snyder drew attention to requirements from JPL which he had presented in the form of four draft TRs (Physical or engineering units, Accessor procedures, Coroutines and Iterators, A more complete type system) together with a set of general proposals; these had been put on the J3 server. He asked that these be considered at the next WG5 meeting. It was possible that JPL would not support his future membership of PL22.3 if a no-development agenda were to be adopted. There was concern over WG5's workload and no decision was reached. 11. Closing Business 11.1 Future meetings The 2011 and 2012 meetings had been dealt with the previous day. Toon Moene offered to investigate the possibility of holding the 2013 meeting in the Netherlands. 11.2 Any other business None was raised. 12. Adoption of Resolutions (N1811) Resolutions LV1, LV2, LV9 and LV10 were approved by unanimous acclaim and LV3, LV4, LV5, LV7 and LV8 by unanimous consent. Resolution LV6 was approved by 13 individual votes to none with one abstention and unanimously by member bodies. 13. Adjournment John Reid thanked the local hosts (PL22.3) again for their excellent support during the week. The meeting closed at 09:10 on Friday, February 19, 2010.