Doc No:   SC22/WG21/N2266
                                                  J16/07-0126
                                        Date:     2007-05-05
                                        Project:  JTC1.22.32
                                        Reply to: Robert Klarer
                                                  IBM Canada, Ltd.
                                                  klarer@ca.ibm.com

Minutes of J16 Meeting No. 44/WG21 Meeting No. 39, April 16-20, 2007

1. Opening activities

Clamage called the meeting to order at 09:10 (GMT) on Monday, April 16, 2007

1.1 Opening comments

Goldthwaite described the arrangements and facilities for the meeting.

1.2 Introductions

Clamage had the attendees introduce themselves.

1.3 Meeting guidelines (Anti-Trust)

Clamage reviewed the patent disclosure rules.

1.4 Membership, voting rights, and procedures for the meeting

Nelson reviewed membership rules. Nelson then circulated the attendance list and membership list. Clamage voting rules and reviewed procedures for the meeting.

1.5 Agenda review and approval

Clamage presented the agenda (document J16/07-0061 = WG21/N2201).

Motion to approve the agenda:

1.6 Distribution of position papers, WG progress reports, WG work plans for the week, and other documents that were not distributed before the meeting.

Each of the Working Group chairs presented their plans for the coming week.

Core Working Group (CWG)

Adamczyk reported that the CWG has a long list of papers to review, and that 13 or 14 papers are nearly ready for formal adoption and integration into the Working Draft.

Library Working Group (LWG)

Hinnant reported that the LWG had "30-odd" papers to review, and that itw was certain that the LWG would not have time at this meeting to address all of them.

Hinnant requested a joint meeting with the Evolution Working Group (EWG) to discuss the desing of a threading API for C++0X.

Hinnant also indicated that TR2 is now considered by LWG to be a low-priority work item: "TR2 is on a back-burner."

Evolution Working Group (EWG)

Stroustrup announced the EWG's plan to start their work at this meeting by prioritizing their work for the ensuing week.

The EWG has nine proposals that are likely to be sent over to CWG at this meeting; each of these proposals will be sent to core with one of its proposers.

Stroustrup expressed his desire to close down EWG after this meeting or the next one.

Stroustrup also observed that the two biggest issues before EWG are Concepts and Garbage Collection. The consensus attitude of the committee is that Concepts have been sufficiently "chewed over" so that no special session is required to discuss them further. There is some desire for a special session on Garbage Collection.

Threading

Boehm indicated that a subgroup of EWG will be holding sessions on threading. Some of the discussions are likely to be very controversial.

1.7 Approval of the minutes of the previous meeting

Motion to approve the minutes (document J16/06-0179 = WG21/N2110)

1.8 Report on the WG21 Sunday meeting

Sutter reports:

Five countries are represented at this meeting; all five have voting status.

Three items have been at ballot since the last meeting:

• the Registration Ballot for the Decimal Arithmetic Types for C++ TR
• the New Work Item Proposal for the C++ Modules TR
• the Registration Ballot for the C++0x Working Draft

The last of these ballots is not yet complete, but this does not prohibit discussion of the draft at this meeting.

The drafting committee will be composed of Adamczyk, Dos Reis, and Hinnant.

1.9 Liaison reports

WG14 Liaison

Plauger reports:

In Portland, WG14 agreed to discuss the question of whether the C standard should be revised at their next meeting, in London.

As well, there will be discussion of some small changes to the Decimal TR in London.

Austern asked whether the Special Math Functions will be made a part of a revised C standard. Plauger responded that that will be discussed by WG14 in London. Austern then observed "that suggests that we should discuss it here this week."

Glassborow noted that "there is always resistance to the prospect of adding new libraries to C."

Sutter suggested that WG21 should discuss the issue at this meeting so that the group can provide liaison to WG14.

Plauger agreed, noting that if C++ agrees to adopt the functions for C++0x, then C is more likely to adopt them, too. Plauger then requested some brief agenda time to discuss this.

Austern also asked for time to discuss this in LWG.

Sutter asked whether the possibility of removing C99 features been discussed?

Plauger explained that "there are very few presumptions going into this meeting."

OWG Vulnerabilities Liaison

OWGV will be meeting in Kona HI, starting Sunday, September 30th in Kona HI and continuing for two more days. WG21 members that want to sit in on OWGV may do so. OWG attendance is not restricted.

TG5 Liaison

Plum explained that this activity has "gone essentially completely inactive."

1.10 Editor's report and WP approval

The current draft of the WP is J16/06-0205 = WG21/N2134. The editor's report corresponding to this draft is J16/06-0204 = WG21/N2133

Motion to accept the working paper (J16/06-0205 = WG21/N2134 ):

1.11 New business requiring actions by the committee

2. Organize subgroups, establish working procedures.

We have three subgroups: Core, Library, and Evolution. There will be a subgroup of Evolution to deal with issues relating to concurrency.

The committee broke into subgroups at 10:30 (GMT).

3. WG sessions (Core, Library, Performance, Evolution).

4. WG sessions continue.

5. WG sessions continue.

6. WG sessions continue.

7. General session.

7.1 WG status and progress reports.

Core Working Group

Adamczyk presented Core Working Group status and reviewed formal motions to be made Friday (for formal motions, see 9.1, below).

EDG will host a "mini-core concepts meeting" in New Jersey.

Motion 1. Move all issues in Ready status in J16/07-0048=WG21/N2188 to DR status and into the Working Draft. Issue numbers are 215, 218, 327, 357, 397, 413, 488, 491, 520, 537, 538, and 582.

Motion 2. Adopt J16/07-0095=WG21/N2235, "Generalized Constant Expression -- Revision 5".

Straw Poll: support for Motion 2

Motion 3. Adopt J16/07-0109=WG21/N2249, "New Character Types in C++".

Straw Poll: support for Motion 3

Motion 4. Adopt J16/07-0118=WG21/N2258, "Template aliases".

Nelson asked whether this proposal had been implemented. Dos Reis explained that it had not.

Straw Poll: support for Motion 4

Motion 5. Adopt J16/07-0099=WG21/N2239, "A finer-grained alternative to sequence points (revised)".

Plum indicated that this proposal is functionally equivalent to the WG14 model.

Vollmann expressed the belief that there is an error of one word in the paper. Nelson disagreed, and promised to work with Vollmann to satisfy his concern.

Sutter asked whether this concern will prompt Switzerland to vote "no" on this item, and Vollmann replied that it would.

Straw Poll: support for Motion 5

Motion 6. Adopt J16/07-0102=WG21/N2242, "Proposed Wording for Variadic Templates (Revision 2)".

Straw Poll: support for Motion 6

Motion 7. Adopt J16/07-0113=WG21/N2253, "Extending sizeof to apply to non-static data members without an object (revision 1)".

Miller explained that this same mechanism will apply to decltype and alignof.

Straw Poll: support for Motion 7

Library Working Group

Hinnant presented Library Working Group status and reviewed formal motions to be made Friday (for formal motions, see 9.1, below).

Motion 0. Move we ask WG14 to treat special math functions in a way consistent with what WG21 is doing.

Plauger noted that the Special Math Functions are currently specified only in a non-normative Technical Report (TR1), and then identified the following options that are available to WG21:

1. the WG can decide to add the Special Math Functions to C++0x, just like the rest of TR1
2. the WG can decide to make a separate IS (normative) that covers "special math"
3. the WG can decide to leave "special math" in TR1.

Reviewing the arguments in favor of Option 1, Plauger noted that three different implementations of this library are available now or are soon to be available, and that these functions meet a need in a sophisticated (but relatively small) community. Plauger also noted that past implementations of these functions have been checkered.

Plauger then reviewed some of the arguments against Option 1, explaining that the implementation of these functions may be too difficult to require of all conforming implementations, and that testing, QA, and customer support will all be costly, even where free implementations are available.

Explaining Option, 2, Plauger observed that these interfaces will have more authority as an IS than as a TR, and that an implementation can conform to C++ without supporting "special math" (or conversely).

Glassborow noted that, if WG21 starts putting specialist libraries into the standard, the standard will soon get very large. The committee should remember that it has the option of issuing a library IS.

Koznik stated a list of objections to Option 1:

• the GNU and Boost implementations are not complete
• partial implementations are demonstrating benefits of slightly different library interfaces than the ones specified in TR1
• testing remains very costly
• support remains very costly
• it will become unclear as to what kinds of functions are too insignificant to be put into the standard
• C++ is better served by a different numerical focus, including real number classes, for example

Plum suggested that the vendors that support the embedded marketplace would appreciate it if we didn't greatly expand the footprint of the library in this way.

Abrahams confirmed that the Boost implementation is not complete.

Meredith reported that the major omission from the Boost and GNU implementations is the hyper-geometric functions, and those were removed from TR1 by a DR.

Sutter spoke in favour of making this library the subject of its own separate IS, noting that this general approach could be a good way of growing the C++ library without raising the spectre of optional parts and that -- besides being a vehicle for special libraries, this could be a useful vehicle for future bindings to other standards.

Discussion ensued.

Austern expressed concern that, if WG21 issues multiple separate International Standards, then the result will be a combinatorial expansion in the number of possible modes of conformance.

Sutter noted that people that care about a binding are already dealing with a combination of multiple standards.

Dos Reis strongly agreed with Austern, and observed that, in the current standard, we already have bindings to other standards, including regex, which was just recently added to the standard.

Spicer suggested that WG21 should consider the suitability of a library to our standard before adopting it into a TR

Crowl spoke in favor of the use of separate IS documents to specify optional pieces of the C++ standard, remarking that this solution is strong enough to meet the requirements of purchasing agents and weak enough to suit the concerns of implementors that are not in the market that is being served by the separate IS.

More discussion ensued.

Straw Poll:Add "special math" to C++0x.

The convenor ruled that this did not constitute consensus.

Straw Poll:Make a new IS for "special math."

The convenor ruled that this constitutes consensus, and that WG21 would send a liaison statement reflecting this consensus to WG14.

Motion 1. Move Ready issues into C++0X Working Paper: 201, 206, 233, 254, 258, 416, 422, 456, 534, 542, 543, 545, 559, 575, 576, 578, 586, 589, 593, 609, 610, 611, 613, 616, 619

Straw Poll: support for Motion 1

Motion 2. Adopt the papers N1856, N1857, N1858, N1859, N18560, N1861, N1862 into the C++0x Working Paper.

• N1856 Rvalue Reference Recommendations for Chapter 20
• N1857 Rvalue Reference Recommendations for Chapter 21
• N1858 Rvalue Reference Recommendations for Chapter 23
• N1859 Rvalue Reference Recommendations for Chapter 24
• N1860 Rvalue Reference Recommendations for Chapter 25
• N1861 Rvalue Reference Recommendations for Chapter 26
• N1862 Rvalue Reference Recommendations for Chapter 27

Straw Poll: support for Motion 2

Motion 3. Move N2255: "Minor Modifications to the type_traits Wording Revision 2" in the C++0X Working Paper. Move N2244: "Wording for decay, make_pair, and make_tuple" into the C++0X Working Paper. Move N2240: "Two missing traits: enable_if and conditional" into the C++0X Working Paper.

Straw Poll: support for Motion 3

Motion 4. Move N2257: "Removing unused allocator functions" into the C++0x Working Paper.

Plauger noted that this isn't just house-cleaning; this change is helpful to the forthcoming "emplace" proposal

Seymour asked whether consideration given to deprecation. Dawes explained that deprecation was considered, but outright removal was preferred because these functions are actively harmful

Straw Poll: support for Motion 4

Motion 5. Move N2246: "2 of the least crazy ideas for the standard library in C++0x" into the C++0x Working Paper.

Straw Poll: support for Motion 5

Motion 6. Move N2241: "Diagnostics Enhancements for C++0x (Rev. 1)" into the C++0X Working Paper.

Straw Poll: support for Motion 6

Motion 7. Move N2238: "Minimal Unicode support for the standard library (revision 3)" into the C++0X Working Paper.

Straw Poll: support for Motion 7

Motion 8. Move N2259: "Specify header dependency for <iostream>" into the C++0X Working Paper.

Straw Poll: support for Motion 8

Motion 9. Move N2192: "Variadic Templates for the C++0x Standard Library (Revision 1)" into the C++0X Working Paper.

Straw Poll: support for Motion 9

Motion 10. Move N2194: "decltype for the C++0x Standard Library" into the second Library Technical Report (TR2).

Motion 11. Adopt N2219: "Constant Expressions in the Standard Library" into the C++0x Working Paper.

Straw Poll: support for Motion 11

Motion 12. Adopt N2211: "Enhancing the time_get facet for POSIX compatibility, Revision 1" into TR2.

Austern observed that the LWG voided changing existing classes in TR1. With this motion, LWG is proposing to change this practice.

Crowl expressed concern about binary incompatibility, as there would be no way for an implementation to conform to the C++ standard and the TR at the same time.

Dos Reis stated that this is an ODR violation.

Sutter asked why this was directed by LWG to TR2 instead of C++0x?

Plauger explained that times are hard to parse, so the LWG believes that the community should have a chance to identify holes in the specification.

Austern spoke against modifying existing library facilities in a TR.

Crowl suggested that the motion be withdrawn.

Sebor reported that this feature has been implemented in the RogueWave library for several years

Crowl indicated that he is unwilling to accept an incompatible change in TR2, but will accept an incompatible change in C++0x.

Dawes and Stroustrup repeated calls to withdraw the motion and resolve the objections in LWG.

The motion was withdrawn.

Evolution Working Group

Evolution Working Group (EWG) will be making no formal motions Friday.

Concurrency Group

The Concurrency Group will be making no formal motions Friday.

Future meetings:

See 10.1, below.

7.2 Presentation and discussion of DRs ready to be voted on. Straw votes taken.

8. WG sessions continue

9. Review of the meeting

9.1 Formal motions, including DRs to be resolved.

Core Working Group Motions

Motion 1. Move all issues in Ready status in J16/07-0048=WG21/N2188 except 488 to DR status and into the Working Draft. Issue numbers are 215, 218, 327, 357, 397, 413, 491, 520, 537, 538, and 582.

Note: The Thursday Straw Poll on this item included issue 488. It was removed for formal voting.

Motion 2. Adopt J16/07-0095=WG21/N2235, "Generalized Constant Expression -- Revision 5".

This motion was withdrawn with then consent of the paper's author, due to the emergence of a technical problem that emerged concerning situations in which overload resolution might have to be performed during template type parameter deduction.

Austern reminded the committee that there is a library motion that depends on this language feature.

Discussion ensued.

Nelson reported that the worst possible outcome of the technical problem in the proposal is that constexpr might not be useful as a non-type template argument.

More discussion ensued.

Stroustrup, in light of the technical debate, supported withdrawal of the motion.

Motion 3. Adopt J16/07-0109=WG21/N2249, "New Character Types in C++".

Motion 4. Adopt J16/07-0118=WG21/N2258, "Template aliases".

Motion 5. Adopt J16/07-0099=WG21/N2239, "A finer-grained alternative to sequence points (revised)".

Straw Poll: support for Motion 5

Motion 6. Adopt J16/07-0102=WG21/N2242, "Proposed Wording for Variadic Templates (Revision 2)".

Motion 7. Adopt J16/07-0113=WG21/N2253, "Extending sizeof to apply to non-static data members without an object (revision 1)".

Library Working Group Motions

Motion 1. Move we empower the Convener to request a New Work Item to prepare an International Standard for the C++ binding of ISO 31 math functions, as described in our Library TR1 (as corrected to date).

Move we inform WG14 of our plans for a special math IS and request that they treat C standardization of special math in a consistent manner.

Motion 2. Move Ready issues into C++0X Working Paper: 201, 206, 233, 254, 258, 416, 422, 456, 534, 542, 543, 545, 559, 575, 576, 578, 586, 589, 593, 609, 610, 611, 613, 616, 619

Motion 3. Adopt the papers N1856, N1857, N1858, N1859, N18560, N1861, N1862 into the C++0x Working Paper.

• N1856 Rvalue Reference Recommendations for Chapter 20
• N1857 Rvalue Reference Recommendations for Chapter 21
• N1858 Rvalue Reference Recommendations for Chapter 23
• N1859 Rvalue Reference Recommendations for Chapter 24
• N1860 Rvalue Reference Recommendations for Chapter 25
• N1861 Rvalue Reference Recommendations for Chapter 26
• N1862 Rvalue Reference Recommendations for Chapter 27

Motion 4.

Move N2255: "Minor Modifications to the type_traits Wording Revision 2" in the C++0X Working Paper.

Move N2244: "Wording for decay, make_pair, and make_tuple" into the C++0X Working Paper.

Move N2240: "Two missing traits: enable_if and conditional" into the C++0X Working Paper.

Motion 5. Move N2257: "Removing unused allocator functions" into the C++0x Working Paper.

Vollmann asked whether this proposal was first introduced at this meeting and, if so, whether there was any reason that formal voting couldn't be delayed until the next meeting.

Hinnant agreed that this proposal was first introduced at this meeting.

Dawes spoke in favor of processing proposals that are ready for formal voting rather than deferring them.

Austern reminded the WG that there is a specific proposal that LWG wishes to move into C++0x at the next meeting that relates to the inplace construction of elements in containers. Accepting this paper makes it possible to provide direction to the author of the other paper.

Sutter suggested to the WG chairs that if they intend to bring something forward at the same meeting at which it was introduced, that they give voting members ample warning.

Rao explained that some companies cannot afford to send members to every meeting, and it would be appreciated if more advance warning were given.

Discussion ensued.

The motion was withdrawn at Hinnant and Austern's request.

Motion 6. Move N2246: "2 of the least crazy ideas for the standard library in C++0x" into the C++0x Working Paper.

Motion 7. Move N2241: "Diagnostics Enhancements for C++0x (Rev. 1)" into the C++0X Working Paper.

Motion 8. Move N2238: "Minimal Unicode support for the standard library (revision 3)" into the C++0X Working Paper.

Motion 9. Move N2259: "Specify header dependency for <iostream>" into the C++0X Working Paper.

Motion 10. Move N2192: "Variadic Templates for the C++0x Standard Library (Revision 1)" into the C++0X Working Paper.

Motion 11. Adopt N2219: "Constant Expressions in the Standard Library" into the C++0x Working Paper.

Evolution Working Group motions

None.

9.2 Review of action items, decisions made, and documents approved by the committee

There are no outstanding action items.

Plum moved to thank the host and all the co-sponsors. Unanimous consent.

Stroustrup presented EWG progress: started off by moving several proposals to core: variadic templates nullptr strongly type enums __func__ alignment support inheriting ctors explicit conversion operators initializer lists PODs defaulted and deleted functions syntactic disambiguation using template concepts new for-loop decltype raw string literals UTF8 literals need more work (before Toronto): programmer-controlled GC names, linkage, and templates (local classes) lambda expressions in-class initializers extensible literal not yet processed: attribute grammar extending rvalue reference to *this being handled in Concurrency subgroup: atomic operations with multi-threaded environments thread-local storage propagating exceptions when joining threads dynamic initialization and concurrency no plans to handle without further input: the rest

Stroustrup: proposal is to leave the Concurrency subgroup alone, and to split the rest of EWG into three; one third could go to LWG, one third could go to CWG, and one third could go to GC and lambda issues.

9.3 Issues delayed until Friday

None.

10. Plans for the future

10.1 Next meeting

Sutter reports:

We had much discussion early in the year concerning the question of whether the July 2007 meeting would be a full meeting or an ad hoc meeting. At the WG21 meeting on Sunday night it was decided that the July meeting would be a full meeting. The penalty for missing this meeting will be waived, for those concerned about maintaining voting rights.

Stoughton announced that the POSIX WG has been considering a new project to define a binding to C++. The study group will hold a meeting from 9am to 5pm on Sunday, July 15 in the meeting hotel.

10.2 Mailings

Nelson reported the following mailing deadlines:

post-meeting mailing	May 4, 2007
pre-Toronto mailing	June 22, 2007

10.3 Following meetings

Plum noted that the information for the next meeting in Kona HI has been in the mailing several times. We will be staying downtown Kona at the Royal Kona Resort. The dates for the meeting are Sunday, September 29 through Saturday, October 6. This will be a six-day meeting.

There will be a liaison meeting of the OWG Vulnerabilities group on Sunday, September 29.

Sutter explained that he is planning to convene three meetings in 2008. The first of these meetings will take place during the week of March 12, in Redmond, and will be hosted by Google. The second of these meetings will take place in late June. Texas A&M universitywill be hosting the meeting in Sofia Antipolis, France. The third meeting in 2008 will take place in November in the San Francisco Bay area, in the Google offices. This meeting will be hosted by Microsoft.

Motion to adjourn

Applause.

Meeting adjourned at 03:20 (GMT)

Attendance