ISO/IEC JTC1/SC22/WG5 N1631 Minutes Meeting of ISO/IEC JTC1/SC22/WG5 May 9th - May 13th, 2005 NEN, Vlinderweg 6, 2623 AX Delft, The Netherlands List of Attendees: John Reid (Convenor) Malcolm Cohen Ian Chivers (Secretary, Friday) Jane Sleightholme Lawrie Schonfelder David Muxworthy Steve Morgan (Secretary, Monday-Thursday) Wolfgang Walter Michael Metcalf (Tuesday) Craig Rasmussen Rob James Dick Hendrickson Dan Nagle Michael Ingrassia Van Snyder William Long Aleksander Donev Masayuki Takata Matthijs van Waveren Toon Moene Kees Ampt (Tuesday p.m.) Nico Vossenstijn (Tuesday p.m.) Arjen Markus (Thurs a.m.) 1. Opening of the Meeting: The meeting opened at 9:00 am, Monday, May 9, 2005 2. Opening business 2.1 Introductory remarks from the Convener The convenor reminded the meeting that this was a joint WG5/J3 meeting with some sessions with John Reid in the chair and some sessions will be of J3 with Dan Nagle in the chair (the exact timings to be announced each day). The main objectives were to be: A. Consider the Repository of Requirements (N1624) and make a preliminary choice of features for the next revision at severity level 5 and above, as defined in N1594. B. Process the interpretations that have completed their J3 letter ballot (see N1612, N1617, N1622) and prepare the first corrigendum for F2003, see N1620 and N1623. 2.2 Welcome from the Host Mr Ton van Bergeijk from NEN welcomed delegates to NEN, Delft and wished the meeting every success. 2.3 Local arrangements Matthijs van Waveren and Toon Moene were making local arrangements. 2.4 Appointments for this meeting David Muxworthy was appointed chair of the Resolutions committee. Steve Morgan was appointed secretary. Matthias Van waveren was appointed Librarian. 2.5 Adoption of the Agenda N1614 accepted with 7a. Editorial Matters (Cohen paper on Enhanced Modules (N1621) to be discussed). 2.6 Approval of the Minutes of the Las Vegas 2004 Meeting [N1598] N1598 approved 3. Matters arising from the minutes None. 4. Status of Las Vegas 2004 Resolutions [N1597] The 2003 standard and the Enhanced Modules TR had now been published. In both cases, the interaction with ISO was poor. The result is that the published versions are not as satisfactory as they might have been, but there are no technical issues. The base for the next revision must be J3/04-007 and not the published standard. 5. Reports 5.1 SC22 Matters (Convener) The convenor had attended SC22 in South Korea. There had been nothing controversial at that meeting concerning Fortran. There was some discussion about language committees adopting a common set of terms. Roger Scowen began creating a Compendium of terms but his ideas were not accepted during follow-up meetings. Rapporteur group established to review Internationalisation (WG20 disbanded) and canvas other WGs for current and future plans and recommend how to proceed. John Reid would provide a report from WG5. WG5's internationalisation policy is in N1619. 5.2 National Activity Reports (Heads of Delegations) Reports from USA (N1627), Japan (N1633, to be written), UK (N1625), Netherlands (N1634), Germany (no paper), Canada (no paper) were given. The convenor noted that France had withdrawn from Fortran work within SC22. 5.3 Report from Primary Development Body (NCITS/J3 Chair) See N1627. 5.4 Reports from other Development Bodies (Editors/Heads) See resolutions. 5.5 Liaison Reports Matthijs van Waveren reported that release 2.5 of OpenMP now supported Fortran 95. The standard now combines coverage of the base languages. 6. Consider the Repository of Requirements (SD5) and make a preliminary choice of significant features for the next revision, that is, features of severity level 5 and above, as defined in N1594. The meeting then split into several subgroups: HPC, Data, JoR were to consider the repository items. Tue p.m. Presentation by Van Snyder on Parameterised modules. Data subgroup recommended that only types and initialisation expressions be allowed as parameters. Straw Vote on form of the Typeless proposal: Fixed sizes as in present proposal 4 Variable sizes 10 Undecided 4 It was then agreed that the 'Typeless' proposal be called 'Bits'. The three large items from the repository, Co-arrays, Bits, and Parameterised Modules were then considered. Following much discussion a straw vote was taken. Straw Vote on the three large items from the Repository: Do none of them 3 Do Co-arrays only 4 Do Bits only 0 Do Parameterised Modules (P.M.) only 0 Do Bits and P.M. 0 Do Co-arrays and P.M. 4 Do Co-arrays and Bits 4 Do all three 4 Undecided 0 Straw Vote: Should we do Co-arrays regardless? Yes 16 No 2 Undecided 1 Straw Vote: Do Co-arrays only 7 Do Co-arrays and Bits 4 Do Coarrays and P.M. 4 Do all three 4 Undecided 0 Other Proposals above Level 4 JoR had looked at 4 issues: RU-004 (Subset for Fortran 2003). The meeting agreed by unanimous consent to discard this item. RU-005 (Extend set of array features). This was deferred. J3-011 (Co-routines). The meeting agreed by unanimous consent not to do this at this time. J3-009 (I/O unit as standard derived type). The data subgroup was split on this. Good basic approach but incomplete as proposed. Straw Vote on J3-009: Do it 5 Nice to do but defer at present 11 Don't do it 1 Undecided 0 J3-043 (Pointers to contiguous memory). The HPC subgroup recommended doing it. Do it 13 Nice to do but defer at present 1 Don't do it 1 Undecided 2 J3-012 (Allocatable & POINTER in generic resolution). Do it 2 Nice to do but defer at present 12 Don't do it 3 Undecided 1 UK-009 (Procedureness in generic resolution) Do it 5 Nice to do but defer at present 12 Don't do it 2 Undecided 0 J3-045 (Same assumed shape declaration). The subgroup felt there were unsafe aspects to this proposal. Do it 1 Nice to do but defer at present 1 Don't do it 15 Undecided 2 Level 4 Proposals J3-019 Math Intrinsics (JoR recommended acceptance) Do it 16 Nice to do but defer at present 0 Don't do it 0 Undecided 1 J3-038 (Bessel, erf,gamma) Do it 6 Nice to do but defer at present 9 Don't do it 2 Undecided 0 J3-017 (Default for optional dummy) Do it 0 Nice to do but defer at present 3 Don't do it 13 Undecided 2 J3-037 (Pointers and targets) Do it 1 Nice to do but defer at present 2 Don't do it 11 Undecided 3 J3-040 (Compute if actual arg is present) Do it 1 Nice to do but defer at present 2 Don't do it 9 Undecided 5 UK-002 (Decimal floating point arith) Do it 9 Nice to do but defer at present 6 Don't do it 0 Undecided 2 UK-003 (Conformance to IEEE 754r) Do it 8 Nice to do but defer at present 9 Don't do it 0 Undecided 1 UK-005 (Long integers) Do it 9 Nice to do but defer at present 2 Don't do it 5 Undecided 2 UK-006 (Multiple non-zero part refs) Do it 5 Nice to do but defer at present 3 Don't do it 3 Undecided 8 UK-010 (Partial initialisation of PARAMETER). Subgroup against. Do it 0 Nice to do but defer at present 4 Don't do it 11 Undecided 4 J3-041 (Interop of pointers, allocatable and assumed shape) Do it 6 Nice to do but defer at present 10 Don't do it 2 Undecided 1 J3-042 (Interop of optional args) Do it 2 Nice to do but defer at present 12 Don't do it 2 Undecided 3 J3-044 (INTENT(SCRATCH)) Do it 10 Nice to do but defer at present 2 Don't do it 5 Undecided 2 JoR items J3-003 (Execute-Command-Line) (Subgroup against) Do it 11 Nice to do but defer at present 5 Don't do it 1 Undecided 2 J3-007 (Construct names local to construct) Do it 1 Nice to do but defer at present 5 Don't do it 8 Undecided 5 J3-031 (ANDTHEN & ORELSE) Do it 1 Nice to do but defer at present 3 Don't do it 11 Undecided 4 J3-036 (USE, EXCEPT) Do it 0 Nice to do but defer at present 5 Don't do it 10 Undecided 4 J3-048 (Write CSV files using LDIO). Subgroup thought 'Nice to do'. Do it 0 Nice to do but defer at present 12 Don't do it 5 Undecided 2 J3-049 (Embed decision in expr.) Do it 2 Nice to do but defer at present 3 Don't do it 10 Undecided 4 Data Subgroup Items ------------------- J3-013 (Internal subprograms as actual args) Do it 13 Nice to do but defer at present 3 Don't do it 0 Undecided 3 J3-015 (Updating COMPLEX parts) Do it 4 Nice to do but defer at present 13 Don't do it 1 Undecided 1 J3-016 (Disassoc/dealloc actual => not Present) Do it 1 Nice to do but defer at present 7 Don't do it 3 Undecided 8 J3-018 (Non-NULL initial targets for pointers) Do it 6 Nice to do but defer at present 9 Don't do it 1 Undecided 3 J3-021 (Resolve generic?) Subgroup thought syntactic cyanide Do it 0 Nice to do but defer at present 0 Don't do it 16 Undecided 3 J3-022 (Polymorphic allocatable in intrinsic assignment) Do it 0 Nice to do but defer at present 15 Don't do it 0 Undecided 4 UK-007 (Pointer function refs as actual args) Do it 10 Nice to do but defer at present 7 Don't do it 0 Undecided 1 UK-008 (Pointer function refs on LHS of assignment) Do it 7 Nice to do but defer at present 5 Don't do it 2 Undecided 3 RU-001 and J3-039 (Maximum rank > 7) Straw Vote: Acceptable rank (multiple votes allowed, must vote at least once) Rank 7 10 15 20 31 Vote 8 10 15 4 4 Straw Vote: combined or separate limit? rank + co-rank <= 15 14 rank only <=15 +extra co-rank 2 No limit 1 Undecided 2 A straw vote was taken on a modified paper which stated that the sum of the normal rank plus the co-rank of an array should be less than or equal to 15. Do it 11 Nice to do but defer at present 2 Don't do it 1 Undecided 3 RU-003 (Extending list of obsolescent features) Straw vote: Do we make a change to the obsolescent features list (add or remove?) Yes 11 No 3 Undecided 2 The convenor then summarised the results of the votes. There were 28 items with a 4 rating. The meeting had said 'yes' to roughly one third, 'no' to roughly one third, and were undecided on roughly one third. There were 6 definite 'yes', 7 definite 'no', 5 were thought to be a good idea of which 2 were probable 'yes' and 1 probably 'no'. Another straw vote was then taken on the three major items in the light of the other straw votes. Straw Vote on the Bits proposal: Do it 4 Nice to do but defer at present 5 Don't do it 8 Undecided 1 Straw Vote on the Generic Modules Proposal Do it 7 Nice to do but defer at present 6 Don't do it 4 Undecided 1 Straw Vote on the Co-arrays proposal: Do it 12 Nice to do but defer at present 2 Don't do it 1 Undecided 2 ================================================= Friday May 13 2005 RU-03 Subgroup recommended adding BLOCK DATA ENTRY DOUBLE PRECISION to the Obsolescent list and Statement Functions to the Deleted list. Several straw votes were taken (Yes/No/Undecided):- Make double precision obsolescent 0-6-7 Delete statement functions 10-3-2 Make .gt., etc obsolescent 1-12-3 Make BLOCK DATA obsolescent UC: no Make ENTRY obsolescent 12-1-3 It was decided to add Statement Functions to the Deleted list and make ENTRY obsolescent (see N1638). RU-05 It was decided to add some of the HPF functions as intrinsics (see N1637). J3-006 and UK-004 A straw vote was taken on which of J3-006 and UK-004 was preferred. The result was 13-0-3 in favour of UK-004. Other papers discussed during the day are recorded in N1635. The portion of the table recording the straw votes taken on items not recorded in the above minutes under agenda item 6 is shown in tabular form below for convenience. In the following: A = Require, B = Allow, C = Decide Later, D = Don't do it, U = Undecided Number Wt * 12 May 2005 ballot 13 May 2005 ballot * = disposition J3-001 3 A A:9 B:4 D:1 U:1 J3-002 3 J3-004 3 C A:1 B:4 D:1 U:9 J3-005 3 C A:1 B:3 D:3 U:9 J3-008 3 B B:uc J3-010 3 A A:11 . . . J3-014 6 B A:7 B:6 D:4 U:1 A:7 B:6 D:4 U:1 Revised by 05-185r1 J3-020 3 A A:9 B:5 D:0 U:1 J3-023 3 B A:7 B:7 D:1 U:1 Revised by 05-189 J3-024 Duplicates RU-002 (see below) J3-025 3 C A:0 B:8 D:4 U:4 J3-026 3 J3-027 3 A A:uc J3-028 3 J3-029 3 J3-030 3 J3-032 3 J3-033 3 J3-034 3 C A:1 B:6 D:0 U:8 J3-035 3 RU-002 3 B A:6 B:8 D:2 U:0 RU-004 >4 D D:uc RU-006 2 NB N1635 states that RU-001 duplicates J3-039 and RU-002 duplicates J3-039. These items were considered together in pairs because the requests were technically very similar. This should have been put the other way round - that J3-039 duplicates RU-001 and J3-039 duplicates RU-002. These items have been requested by the Russian Fortran Working Group for many years, see WG5-1189, items 24, 25 and WG5-919, items 1, 2; and before that by the Soviet Fortran Specialist Group, see WG5-713, items 7, 9. o EDIT report N1621 was discussed. A revised version will be produced. Since WG5 had not delegated the Extended Modules TR to J3, discussion of editorial changes to this would wait for next WG5 meeting. ================================================== 7. Consider the Fortran defect reports (interpretations) in J3-006 and, if appropriate, prepare the first Corrigendum for Fortran 2003. N1636 technical corrigendum was agreed. Some minor changes will be needed for eventual publication (see the note at the top of N1636). Many of the outstanding interpretations were discussed and papers will be prepared for the next J3 meeting based on these discussions. 8. WG5 Business and Strategic Plans 8.1 Goals for 2005-2008 To be discussed by email. 9. Closing Business 9.1 Future meetings The next meeting will be in February 2006 in USA, probably at George Mason University, near Washington. The 2007 meeting will be in the UK in July or August, probably in London. 9.2 Any other business None. 10. Adoption of Resolutions The resolutions as recorded in N1630 were all passed by unanimous consent, but note that there are two typos: D6, line 9. Change 'j3/05-185r1' to 'J3/05-181r1'. D6, line 12. Change 'WG5-1637' to 'WG5-N1637'. 11. Adjournment: 1800, May 13, 2005