Draft Minutes for Apr 10-13 2000 Meeting of ISO/JTC1/SC22/WG14 and NCITS/J11 WG14/N914 10 Apr 00 09:30-12:00 13:30-17:00 11 Apr 00 09:00-12:00 13:30-17:00 12 Apr 00 09:00-12:00 13:30-16:30 13 Apr 00 09:00-09:50 Mayumi Mita Secretary General IPSJ/ITSCJ Room 68 Kikai-Shinko-Kaikan Bldg. 3-5-8, Shiba-koen, Minato-ku Tokyo 105-0011 Japan E-mail: mita@itscj.ipsj.or.jp Fax: + 81 3 3431 6493 Phone: + 81 3 3431 2808 ------------------------------------------------------------ Monday April 10th 1. Opening activities 1.1 Opening Comments Benito welcomed everybody to Japan and introduced Japanese standards, our hosts. 1.2 Introduction of Participants/Roll Call Rich Peterson Compaq/Digital Frank Farance Farance Inc Raymond Mak IBM Corp Canada HOD David Keaton David Keaton Randy Meyers Randy Meyers Jeff Muller Oracle John Benito Perennial Convener Meeting chairman Thomas Plum Plum Hall Larry Jones SDRC Douglas Walls Sun Microsystems USA HOD Fred Tydeman Fred Tydeman Jan Kristoffersen Denmark HOD Makoto Noda Japan Randy Marques Orgin Technical Automation, Netherlands Willem Wakker Netherlands HOD Francis Glassborow UK Hiroshi Fukutomi Japan HOD Liam Fitzpatrick Ireland HOD Derek Jones Knowledge Software UK HOD Meeting secretary Tanaka Keishiro Japan 1.3 Selection of Meeting Chair John Benito was selected as meeting chair. 1.4 Procedures for this Meeting Business as usual. 1.5 Approval of Previous Minutes (WG14/N904) The previous minutes were accepted without change. 1.6 Review of Action Items and Resolutions Open: Gwyn to promulgate to newsgroups the DR procedures. Withdrawn: Benito to collect DR's into a TC and forward to SC22 Open: Benito to work with Keld to set up reflector for embedded C. Ongoing: Tydeman to find words from NCEG TR on signalling NaNs to consider for inclusion in Rationale in place of two-line reference to the TR (which is not available). Closed: Mak to find example on restricted pointers with multiple levels of indirection. Closed: Peterson to look into Compaq being co-sponsor. Open: Feather to write rationale for N901 item 2, assignment of partially initialized struct. 1.7 Approval of Agenda (WG14/N911) Correction: Item 4 N910, should refer to N909. Two additional items for the J11 agenda: J11 Tag vote on withdrawing the NCEG report. Form US position on NWI. 1.8 Distribution of New Documents There were no new documents for distribution. 1.9 Information on Next Meeting Information on the next meeting, in Toronto, was included in the mailing for the current meeting. 1.10 Identification of National Bodys National bodies present were: Ireland, Netherlands, Japan, UK, Denmark, Canada, and the US. 1.11 Identification of J11 voting members There are currently 16 voting members of J11, 10 voting members were present at the meeting. 2. Rational Editor's Report [N___] (Seymour) Seymour was not responding to email/phone. The committee was not in possession of the latest version of the rationale document. It was agreed that the rationale needed to be a living document that grew over time. ACTION Keaton: contact rationale editor. 3. NWI Status Report (Benito) There were 7 countries expressing an interest in actively participating. The JTC1 ballot closed on Jun 6. 6. Liaison Activities 6.1 J11 It had been found that no J11 work items have been retired. This situation would be corrected. Voting on new chair closes on Apr 12. Randy Meyers is standing for chair and Fred Tydeman as vice chair. 6.2 WG14 Liaison to ANSI. There has been a protest comment on the ISO standard being adopted as an ANSI standard. ANSI are holding a ballot to resolve the issue. 6.3 J16/WG21 A TC is planned, 2.5 year correction, that will include the record of responses to DRs. The schedule calls for a vote on submitting the TC at the Toronto meeting of WG21. The performance group were continuing to work on their document, but no time scale had been given for when the document would be published. 6.4 WG15 (POSIX) There was some suggestions that the OpenGroup had incorporated all of the C99 document in one of their standards. ACTION Benito: Find out if the OpenGroup had incorporated the text of the C99 standard in one of their standards. 6.5 WG20 (I18N) No liaison. 6.6 Other Liaison Activities Wakker reported that WG11 met a few weeks ago. They have started work on complex data types and complex elementary functions. They were about to start revision of the Language Independent Datatype standard. The document dealing with elementary numerical procedures was being submitted for FDIS. 7. Sequence points and related issues N908 (Jones, UK) A revised N908 had been distributed in the pre-Tokyo mailing. There was some discussion over what the UK wanted to do with this issue. Should any published document be normative or informative? How would it impact existing implementations? Was it planned to cover all expressions? ACTION Jones(UK): submit proposed NWI wording for Wednesday. 5. Performance critical systems N910 (Kristoffersen) Kristoffersen presented paper N910. Many people pointed out that the proposals being presented were trying to get over quality of implementation issues, ie code optimization. It was felt that the way forward was to put pressure on vendors to improve the quality of generated code, rather than have them implement new language features. ACTION Meyers: Create wording for rationale dealing with recommended practice on inline and quality of implementation of optimizations. 4. C++ TC on Basic I/O Hardware addressing N909 (Kristoffersen) Kristoffersen presented paper on work going on in WG21 with I/O hardware addressing. 8. Embedded processor support N907 (Wakker) Willem Wakker presented the fixed point discussion items given in N907. Tuesday April 11th 8a. Continue discussion on N907. Willem Wakker presented the saturation and circular buffer discussion given in N907. There was discussion on how fixed point datatypes might participate in the usual arithmetic conversions. Wakker asked the committee for an indication on how he might proceed with selecting the various parameters discussed in N907. The committee wanted more information on how these constructs were used in real life. ACTION Wakker: Provide references to information on processors that implement fixed point datatypes, saturation arithmetic and circular buffers. ACTION Wakker: Update N907 in light of the committee discussion. The issue of I/O hardware and the DSP NWI was raised. There was a lot of discussion of the UDI work (Uniform Device Driver, www.project-udi.org) and whether the Danish proposals might best be handled by this group. Plum described his understanding, having attended UDI meetings. UDI operates at a much higher level than the existing I/O proposal and the UDI committee might not welcome further liaisons with other standards committees. Some interest was expressed in producing a C binding to UDI. Straw vote: Do you support adding I/O hardware support to the DSP work item? J11 3/4/3 WG14 3/2/2 ACTION Japan, Ireland: Consult with their members on the I/O hardware issue. ACTION Benito: Collate responses from Japan and Ireland. 9. Defect Reports The outstanding DRs were divided up and handed out to three sub groups for them to work on. 205, 206, 210, 211, 212: subgroup chair Fred Tydeman 214-218: subgroup chair Randy Meyers 219-222: subgroup chair Frank Farance Wednesday April 12th 9b. Continue Defect Reports DR219: the subgroup had agreed that the effective type for the cases under consideration was array of char. Lots of discussion followed. What was the intent of effective type? Did it apply to objects only? There were existing implementations that used explicit casts in deducing an based effective type. Straw vote: Is the effective type for the cases in DR219 array of char? 5/7/5 ACTION Meyers: Write paper on why he thinks the effective type of p1 is int (DR219). DR220: Open on points 1 & 2. Other items committee agreed with subgroup wording 13/2/2 DR221: committee agreed with subgroup wording LOTS/0/0 DR222: Lots of discussion. Was the existing worded correct? Straw vote: Existing wording in standard specifies defined behaviour 0/LOTS/0. Straw vote: Change the existing wording to make the behaviour undefined 13/2/2. ACTION Jones(US): Produce words that would change the existing standard words to become undefined behaviour. DR205: Straw vote: do nothing 7, do something 9. DR206: committee agree with subgroup wording LOTS/0/0 DR210: committee agreed with subgroup wording and thought it should be adopted as a TC LOTS/0/1. DR211: Still open. DR212: Still open. DR214: Still open. DR215: committee agreed that 6.5.8p? should be duplicated in 6.5.9 LOTS/0/1. DR216: ACTION Meyers: Produce rationale wording for #if. DR217: Lots of discussion. Straw vote: Leave standard alone 8 allow return to be null & correct example 4 don't allow null to be returned & correct example 3 abstain 2 DR218: committee agreed with subgroup wording. Wording should be adopted as a TC LOTS/0/1. Revisted DSP NWI issues John Benito had received a reply from Keld. Norway supported the I/O proposal being part of the DSP NWI. ACTION Simmonsen: Set up reflector to discuss embedded systems. The reflector will be called embedded_c. ACTION All: Email keld@dkuug.dk if they want to be added to the embedded_c reflector. Miscellaneous discussion There was some discussion on adding recommended practices to the library functions. ACTION Plum: To create an inventory of situations in the library that cause undefined behaviour. Benito announced that he had accepted new DRs DR223-230. He than quickly reviewed these new DRs. Benito announced that Japan had asked for two issues to be accepted as DRs, and he had assigned DR numbers to them. Revisted DR(static). Stepping back from the immediate issue raised, the committee discussed what might be done with the prototype static functionality. The entire construct could be marked as obsolescent, or restrictions could be placed on keyword ordering. What were the politics of removing a feature that has only just be added to the standard? Straw vote: Remove use of static from array types in prototypes? Committee as a whole 7/4/6 J11 4/3/3 WG14 2/0/4 Straw vote: DR(atexit), accept Meyers words LOTS/0/0 Jones(UK) presented draft for a NWI on sequence point issues, N913. Some discussion on the issues involved. It was important to be compatible with C++, or at least a common subset. Straw vote: Who would support work on N913? J11 5/0/5 WG14 2/0/4 A NWI requires support from five National Bodies. Plum asked if the DR machinery was more appropriate for this work. People queried if the amount of work involved was too high for the DR machinery. There were still issues to be resolved in N908. In particular the floating point model needed a lot more work. J11/U.S. TAG meetings held. Thursday April 13th 14. Separate WG14 admin No admin business 16. Administration 16.1 Future Meetings 16.1.1 Future Meeting Schedule Toronto October 16-20, hosted by IBM. Copenhagen April 23-27 2001, hosted by Danish Standards. No host for October 2001 meeting. East coast US suggested. Benito offered Santa Cruz if no other hosts came forward. October 15-19 2001 was suggested for the meeting week. 16.1.2 Future Agenda Items No new items. 16.1.3 Future Mailings Closing date for Post Tokyo mailing will be May 12. Closing date for Pre Toronto mailing September 15. Convenor would like all DR wording by the date of the Post Tokyo mailing. The convenor has not been producing paper mailings for the last year. Email only. ACTION Meyers: Find out if NCITS still require a paper mailing. 16.2 Resolutions 16.2.1 Review of Decisions Reached No formal decisions reached. 16.2.2 Formal Vote on Resolutions No formal votes taken. 16.2.3 Review of Action Items ACTION subgroup chairs: Get DR wording to Benito for Post Tokyo mailing. 16.2.4 Thanks to Host The committee thanked Japan for being a very hospitable host. 16.3 Other Business None. 17. Adjournment The meeting was adjourned at 09:50. derek -- Derek M Jones tel: +44 (0) 1252 520 667 Knowledge Software Ltd mailto:derek@knosof.co.uk Applications Standards Conformance Testing http://www.knosof.co.uk