ISO/IEC JTC1/SC22/WG5/N1654 Minutes Meeting of ISO/IEC JTC1/SC22/WG5 February 13th - February 17th, 2006 George Mason University (GMU), Fairfax, Virginia, 22032, USA List of Attendees: WG5: John Reid (Convenor) Dan Nagle Rich Bleikamp Ian Chivers Malcolm Cohen Aleksander Donev Dick Hendrickson Michael Ingrassia Rob James Bill Long Jeanne Martin David Muxworthy Mallory North Craig Rasmussen Jane Sleightholme (Secretary) Van Snyder Masayuki Takata Matthijs van Waveren (Thursday - Friday) Wolfgang Walter Stan Whitlock Guests: Glen Luecke (Monday) Iowa State University Bob Numrich (Monday - Thursday) University of Minnesota John Wallin (host) George Mason University Joaquin Arteaga (Friday) George Mason University 1. Opening of the meeting: The meeting officially opened at 9.00am Tuesday 14th February 2006. The convenor John Reid was held up due to bad weather and so the Monday session was J3 business only. John apologised for his late arrival. 2. Opening business 2.1 Introductory remarks from the Convenor The convenor reminded the meeting that this was a joint WG5/ J3 meeting where WG5 sessions would be chaired by John Reid and J3 sessions chaired by Dan Nagle. The main purpose of the meeting was to firm up the requirements for Fortran 2008 confirming our intention to stick to the schedule in N1590. WG5 will not be processing any interpretations at this meeting. After this meeting, there will be a WG5 ballot on the interpretations that are ready to become Fortran 2003 Corrigendum 2. The main objective was: To consider the Repository of Requirements (SD5) and make a firm choice of significant features for the next revision, that is, features of severity level 4 and above, as defined in N1594. 2.2 Welcome from the host Professor John Wallin from George Mason University had welcomed us at the J3 meeting the previous day and wished us every success. 2.3 Local arrangements At the J3 meeting the day before Dan Nagle and Mallory North explained that they were responsible for local arrangements and told us about a trip to the Air and Space Museum at Dulles Airport planned for Wednesday afternoon. The coach would leave GMU at 12pm and return for approximately 4.30 pm. 2.4 Appointments for this meeting David Muxworthy was appointed Chair of the Resolutions committee. Matthijs van Waveren, David Muxworthy, Masayuki Takata, Wolfgang Walter, Rob James, and Van Snyder to be on the Drafting Committee representing all the countries. Jane Sleightholme was appointed secretary. John Reid was appointed Librarian. 2.5 Adoption of the Agenda N1644 accepted. 2.6 Approval of the minutes of the Delft 2005 meeting [N1631] N1631 approved. 3. Matters arising from the minutes None. 4. Status of Delft 2005 Resolutions [N1630] All of the Delft 2005 Resolutions in N1630 were completed. 5. Reports 5.1 SC22 Matters (Convenor) The convenor had attended SC22 in Canada September 2005. a) There had been a resolution asking ISO to make language standards free and more visible. b) John Reid had put in a request that the WG5 TRs be freely available and this has gone out to a vote. c) There is a delay with IEEE 754r: decimal floating-point support. Cobol likes the encoding of 3 decimal digits in 10 bits while others like an encoding that involves long integers representing the exponent and mantissa. John says that Fortran should support decimal floating-point but he doesn't think Fortran cares about the encoding. There was a J3 resolution, passed at meeting 174 by unanimous consent, that stated J3's support for decimal floating-point and deferred to Cobol about the format. d) 05-06 resolution wanting a better procedure in future for the final stages of processing a standard has gone to ISO. e) John Reid has made a formal complaint about the slow processing of standards. For example Van's TR sat in Geneva for 6 months. The whole process should take no longer than 3 years. f) John Reid re-appointed as Convenor for 3 years and Malcolm Cohen appointed as editor. 5.2 National Activity Reports (Heads of Delegations) Report from UK (paper N1651), USA (N1652), Canada (no paper). Germany (verbal report): German group inactive. Japan (N1655 to be written). Netherlands (Matthijs to arrive on Thursday) 5.3 Report from Primary Development Body (NCITS/ J3 Chair). This item can be removed from the agenda. 5.4 Reports from other Development Bodies (Editors/Heads). There are no other development bodies. 5.5 Liaison Reports NCITS/H2 (SQL): Dan Nagle. Nothing that affects Fortran. NCITS/J11 (C): Craig Rasmussen. There have been some discussions about RESTRICT and that needs to be discussed here by subgroup. About DARPA, the C group does not want to define what happens in error conditions. C++ prefers that users write their own exception handler. IEEE/754: Michael Ingrassia. Nothing that affects Fortran; they are trying to decide what to do next. IFIP/WG5: Van Snyder. Nothing to report. 6. Consider the Repository of Requirements (N1649) and make a firm choice of significant features for the next revision, that is, features of severity level 4 and above, as defined in N1594. The meeting voted on the following J3 papers that were assigned to sub-group JoR on Monday: J3/06-110 (A HERE I AM subroutine). Straw vote: Should this be worked on for Fortran 2008? Yes 3 No 11 Undecided 5 J3-025 (Allow SUBROUTINE name or FUNCTION name to be optional on END statements for module and internal subprograms). Straw vote: Should we be doing this as a Fortran 2008 task? Yes 10 No 1 Undecided 6 J3-026 (ATAN with two arguments works like ATAN2). Straw vote: Should it be moved to B list? Yes 18 No 0 Undecided 1 J3/06-117 (Extend INT to LOGICAL argument). A new item not in the Repository. Agreed that it was level 3. Straw vote: Should we add it to the Repository? Yes 2 No 9 Undecided 8 Items in the repository: J3-008 (Rewrite attribute requirements). According to N1635 it's on the B list Straw vote: Move it to the A list? Yes 13 No 3 Don't 0 Undecided 1 Result: It moves to the A list. J3-009 (IO_UNIT standard derived type). Do it 2 Do it if time 1 Don't do it 9 Undecided 7 Result: don't do it. UK-003 (Conformance to IEEE 754R) Straw vote: Do it 0 Do it if time 0 Don't do it 13 Undecided 6 Result: don't do it. J3-002 (GET_IO_UNIT). This is in limbo Straw vote: with the understanding that an alternative using the OPEN statement would be acceptable. Do it 13 Do it if there's time 6 Don't do it 0 Undecided 0 Result: do it. UK-012 (Recursive I/O). This is a new item and needs an ABCD vote. UK put it in at level 3. Straw vote: Do it 8 Do it if time 8 Don't do it 0 Undecided 3 Result: do it if there's time, so it goes on the B list. Repository items in limbo: J3-004 ((STORAGE_SIZE). Straw vote: are we sympathetic to the idea of getting storage size? Yes 14 No 0 Undecided 5 Result: it goes back to JoR J3-032 (Findloc). Vote: Do it 5 Do it if time 12 Don't do it 1 Undecided 1 Result: B J3-035 (Proposed f2k+ MTE on semicolons) Do it 4 Do it if time 13 Don't do it 0 Undecided 2 Result: B J3-033 (Compiler version). This needs an ABD vote. Three-way straw vote on whether we like the idea: Like 13 Don't like 2 Undecided 5 Result - back to JoR for fleshing out and then come back for an ABD vote. Data subgroup items: J3/06-112 J3-016 (Disassociated or deallocated actual argument associated with nonpointer nonallocatable optional dummy argument). Straw vote: Do it 0 Do it if time 7 Don't do it 7 Undecided 4 Result will retake vote in a few days time. J3/06-115 / J3-028 (Allow forward type for allocatable components). Straw vote: Do it 7 Do it if time 10 Don't do it 0 Undecided 1 Result: B (do it if time) J3/06-123 (Intelligent Macros - Specs and Syntax). Straw vote: Do we like the change from parameterised modules to intelligent macros? Yes 15 No 2 Undecided 2 Van raised 2 items a)J3-044 (New Intents). Straw vote on whether this is moved to the Don't Do list: Yes 14 No 0 Undecided 5 Result: D(don't do it) b)RU-003 (Extend the obsolescent features list). Straw vote: Delete 1st item? Yes 12 No 4 Undecided 2 Result: 1st item will be removed. Wed 15th February 9am JoR subgroup reported back to full meeting and the following votes were taken: J3/06-136 from Tuesday folder on server. (Compiler Version). It is specs and syntax for J3-033 ABD vote: A(must do) 2 B(do if time) 13 Don't do 1 Undecided 2 Result: it goes on the B list J3-029 (More information about GET_COMMAND[ ARGUMENT] failure). JoR recommends a D vote. ABD vote: A(must do) 0 B(do if time) 1 Don't do 15 Undecided 2 Result: it goes on the D list RU-005 (Extend a set of array intrinsic functions) JoR recommends D Straw vote: to remove PREFIX and SUFFIX intrinsics Yes 16 No 0 Undecided 2 Data subgroup: J3-030 (Simplified means to select the most commonly desired real and integer kinds) subgroup undecided. ABD vote: A(must do) 2 B(do if time) 10 D(don't do) 6 Undecided 0 Result: it goes on the B list J3-034 (Add MOLD to ALLOCATE) subgroup undecided Like/ don't like vote: Must do 0 Do if time 9 Don't do it 1 Undecided 8 Result: it goes on the B list UK-004 (Kind environment specification) and J3-006 (Find all available logical and character kinds) (UK-004 is a superset of J3-006) Subgroup recommends not doing it. ABD vote: Must do 1 Do if time 3 Don't do 11 Undecided 3 Result: both rejected UK-006 (Multiple nonzero-rank part references) Subgroup dislikes this - lots of edits needed if we go for it. ABD vote: Must do 1 Do if time 0 Don't do 13 Undecided 4 Result: goes on reject list. UK-011 (Elemental procedures that aren't pure) Subgroup recommend doing it. Must do 7 Do if time 10 Don't do 0 Undecided 1 Result: it goes on the B list. /Edit subgroup report: RU-006 (Give a table with attribute compatibility) Result: to become a D. HPC subgroup report: J3-005 (C_SIZEOF) Subgroup recommend doing it. Must do 2 Do if time 11 Don't do 0 Undecided 5 Result: it goes on the B list. Thursday 16th February J3-016 (Disassociated or deallocated actual argument associated with nonpointer nonallocatable optional dummy argument). Retake vote: ABD vote: Must do 2 Do if time 9 Don't do it 3 Undecided 5 Result: B Co-arrays: Bill Long introduced the discussion on co-arrays. I/O Straw votes: Should Rewind/Backspace/ End file be removed? Yes 15 No 0 Undecided 4 Should we keep direct access capabilities? (John Reid said vote yes if you think something should be done but don't know details) Yes 7 No 0 Undecided 11 Straw vote: Should the collectives be kept? Keep collectives 6 Reduce them to sum, min, max 2 Remove them 2 Undecided 8 Friday 17th February JoR subgroup report: J3/06-147 (STORAGE_SIZE) ABD vote: A(Do) 8 B(Do if time) 8 C(Don't do) 0 D(Undecided) 2 Result: B Continuing item 6 on the agenda. Convenor said that we had been doing ABCD votes but we've still got two big items which are Bs at level 6: J3-014 Intelligent macros J3-047 BITS BITS - is further along but will have a bigger impact on the Standard. Macros - probably about the same amount of work required to finish it. Co-arrays - there may be more still not done compared with the others. Straw/straw vote: 4-way vote and can vote more than once. Keep macros & bits 6 Do macros only 7 Do bits only 0 Do neither 4 Result: inconclusive The meeting decided to keep UK-001 "Co-array Fortran for parallel programming" on the A list and to keep J3-014 "Intelligent macros" and J3-047 "BITS" on the B list. J3 will keep the latest draft for Co-array edits on the J3 server under meeting no. 176. 7. Consider the Fortran defect reports (interpretations) in J3-006 and, if appropriate, prepare the second Corrigendum for Fortran 2003. After this meeting there will be a WG5 ballot on the Fortran interpretations that are ready to become Fortran 2003 Corrigendum 2. Stan Whitlock will send that subset of 06-006Ar1 to John Reid. Corrigendum 1 will be published soon. 8. WG5 Business and Strategic Plans 8.1 Goals for 2006-2009 These will wait until after the next J3 meeting because John Reid's report to SC22 is in September 2006. 9. Closing business 9.1 Future meetings - the 2007 meeting will be 6-10 August 2007, London, UK. Hosts Ian Chivers and Jane Sleightholme. 9.2 Any other business - none. 10. Adoption of Resolutions The "resolutions of the WG5 meeting, 13-17 February, 2006, Fairfax, VA, USA" are in N1653: resolutions F1 - F7 were adopted by unanimous consent. Resolutions F8 - F9 were adopted unanimously and we wished to express our thanks to the GMU, and to John and Dan for all that they've done. 11. Meeting adjourned at 16.03, 17 February 2006.