ISO/IEC JTC1 SC22 WG21 N4052 - 2014-06-10
Ville Voutilainen, [email protected]
Teleconference information: Date: 2014-06-06 Time: 8:00am N.Am. Pacific Time Duration: 2 hours
Sutter called the meeting to order 8:10 Pacific Time.
In attendance were:
The agenda N3992 was adopted by unanimous consent.
The minutes in N3900 were adopted by unanimous consent.
There were no action items. Sutter to make a note to include the link to previous minutes in the agenda.
Sutter suggested looking at isocpp.org/std/status for the current editor assignments. Sutter also mentioned that the business plan and convener's report is coming out in June.
Miller reported that there are 66 new Core issues, and 7 pulled from ready to review. He also pointed out that there's no longer a separate protected version in mailings but the committee version is on the Rapperwil meeting wiki.
Since there's an ongoing ballot, Core hasn't been doing drafting reviews between meetings, and will have no motions in Rapperswil. Core did arrange a drafting review teleconference for the Transactional Memory wording by Maurer.
Miller reported that Core has no unusual plans, plan is to do Core issue processing and TS review as needed in Rapperswil. Concepts wording review may appear on the agenda. Concepts lite was already wording-reviewed in Issaquah, but there haven't been any further reviews and no teleconferences between meetings. Sutter reminded that while the main standard Working Draft will not be modified by any motions in Rapperswil, everything else is fair game and business as usual.
LWG plans to continue Filesystem TS ballot resolution and to handle any issues with the Fundamentals TS. The plan is to get the Fundamentals TS out for a PDTS ballot from Rapperswil.
Majority of incoming papers are expected to go to LEWG first, but there have been some LWG papers incoming. Meredith requested people to participate in the tentative voting for the Filesystem TS issues. The plan is to get the Filesystem TS to the DTS balloting step.
Clow pointed out that there are some LEWG/LWG issues that should be clarified direction-wise. Meredith said issues will be processed as usual, but not put beyond Ready status. Sutter proposed starting the meeting with a triage.
Voutilainen reported that there are 86 open EWG issues, and a dozen new papers in the pre-meeting mailing, and four papers that are new revisions of previous proposals.
Voutilainen explained that the large amount of open issues is caused by the EWG issue list tracking Core issues with Extension status, and expected the agenda to consist of incoming papers and issue processing, in addition to TS discussions including at least the Arrays TS and potentially Concepts.
Yasskin said there are 24 topics to discussed, based on the issue count in LEWG Bugzilla. He said that these are mostly TS material, and plans to perform aggressive triaging. He also mentioned the plan to have email threads on the incoming papers on the LEWG reflector.
The plan is to get the Fundamentals TS out for ballot, and review papers incoming for Fundamentals TS v2.
Sutter pointed to the isocpp.org status tracking page. It's not official, but it is indicative. 5 TS documents are planned to go out for PDTS ballot, 1 for DTS ballot.
Dawes said he doesn't see any reason not be ready to ship for DTS ballot, and said that the remaining issues are not contentious. There are 15 issues to be voted on the reflector for Tentatively Ready, and the details are in the published issue lists. Dawes pointed out that a new Microsoft VS CTP is now out with filesystem v3, which is basically the TS, so there are now two implementations of the TS available.
Crowl reported that the Arrays TS has a document and some papers, but what's missing is consensus that the approach is right. Crowl reported that there haven't been changes to the core wording, and thought that there's no need for a review. Miller said there are some core issues related to the wording. Sutter reminded that new proposals need to go through CWG/LWG wording review.
Meredith expressed doubt that Arrays is ready after Rapperswil, and said LEWG apparently wants to add new functionality to the TS, and there are some open issues. Meredith estimated that Urbana is a more likely target. Sutter requested Crowl to coordinate with Meredith and Yasskin, and suggested that to be done early in the meeting. Sutter warned against delaying TSes to 'add one more feature' if the feature could really wait until TS v2, so as to avoid the antipattern that contributed to delaying C++0x for years, and made the remark that increasing cohesion is ok.
Boehm said there's 16 proposals for the Concurrency TS and 3 for the Parallelism TS. He has hopes that the Parallelism TS can be made ready for PDTS ballot, but mentioned that there's no working draft currently.
Boehm reported that there's a lot of churn in the Concurrency TS. He expects to set a cut-off point for TS features, but explained that there's a lot of uncertainty about what the initial content will be. Meredith said there are some doubts about the state of Parallelism and the ability to ship a PDTS. Boehm said Parallelism gets higher priority.
Vollman asked about issues against the TSes. Sutter said that in general there should be an issue list for all the documents, and recommended asking the project editors. Sutter said he'll take an action to try and ensure that all documents have issue lists of some form. Meredith expressed desire to have a list that can be voted on, instead of a changing document. Carruth said the bugzilla list has the sufficient capability to do that, and works as a basis for an N-paper if need be. Meredith clarified that he prefers seeing a paper that has the exact wording that is voted on.
Wong summarized the wording review teleconference, there was 6-7 people, no large issues, planning a second review in Core in Rapperswil. Wong thinks the wording is in good shape for a new Working Draft. Wong expects full TS to be available before Urbana, and the PDTS ballot to go out from Urbana.
Wong said SG5 is continuing with library facilities, like std::list, and requested LWG review. Some language issues such as optimizations are still being processed. Meredith asked what the scope of the TS is, and expressed concern about schedule if library parts are included. Wong said he expects std::list to be included in the TS. Meredith explained that the concern is the complexity of the library changes. Wong said the scope is limited in order to keep it maintainable, and the plan is to incrementally add library components to future TSes.
Wong explained that library components like string and shared_ptr have challenges that take time to deal with. Wong reported he's working with library vendors, and says there's possible compiler implementation work for clang coming up. Meredith and Wong agreed that there needs to be a discussion about the proper scope of the initial and future content of the TS.
Sutter summarized the status of concepts, and said there's been no wording reviews between meetings, that the wording will be adopted as an initial working paper, and that it's unknown whether it's going to be ready for PDTS ballot. Sutter will check the status. Miller said he can schedule wording review if needed.
Gregor reported that there's one design paper to discuss, and that SG2 plans to meet. Clow said that there's been some recent (clang) implementation work ongoing, Suttern recommended contacting the vendors offline for status updates.
Sutter summarized that there's been some incoming papers for SG4, but otherwise the status is unclear.
Crowl reported that there's one new proposal and one revision of a previous propsal, but no plan to meet, since principals (incl. Crowl himself) won't be present, and SG6 will resume in Urbana.
Crowl said Arrays can progress because it's mostly Evolution. Crowl said he tries to be available for communication. Crowl said there's going to be a floating point presentation that's going to target Evolution. Voutilainen said he can help with getting that on the EWG agenda.
Wong asked where Decimal Floating point lands in the C++ ecosystem. Sutter said there's a separate standard that is done and is not active. Meredith said there's an update to that, and there have been requests to get those updates on the agenda. Sutter suggested that if a study group is unavailable, the Evolution groups could be used instead. N3871 is likely to be put on the agenda, Meredith thinks it's a pure library extension, Crowl to coordinate with Yasskin on that.
Carruth reported that there are incoming papers, and is planning for an evening session. One paper almost ready for (L)EWG review, going to prioritize, potential plans for initial TS content. The remainder of the agenda is to review papers and provide feedback to authors.
Clow reported that there are no papers for Rapperswil. Clow said he plans to encourage people to write papers, and Carruth reported that Niebler has a paper for Urbana and wants to be present for the paper, but can't be in Rapperswil.
Nelson said there's a new paper, and there's also a proposal for has_attribute, says there's a need for SG10/EWG/LEWG cooperation.
Sutter reported that SG11 is inactive, there are some database papers, which are for the time being going to LEWG.
Dos Reis reported there are two papers dealing with UB in preprocessor, planning to have an evening session in Rapperswil.
Sutter reported that there's an updated paper, presentation will be done in a session of some kind.
Sutter summarized the JTC1 directives changes: there are no longer NB delegations, but individual experts only. No heads of delegation, changing voting to every individual in the room who is in the global directory. Sutter explained that he still plans to query the NBs separately for any major objections as a sanity check.
van Winkel asked how Sutter would know which NB the experts represent. Sutter explained that the global directory designates the NB for every individual expert. Sutter said he's not sure what happens if an individual is an expert of multiple NBs. Wong said it's possible to be just one NB at the time Global Directory -wise.
Wong requested an explanation of the motivation of this change. Sutter said these changes come from the JTC1 level from the national bodies, and the procedural change comes from there. Plum said that JTC1 has worked differently from ISO previously. Goldthwaite speculated that the success of WG21 may be a reason for these changes and the increased involvement of JTC1/SC22 in the work of the Working Groups.
Vollman said he's in the ISO directory as an expert member of the Swiss NB, but not necessarily as a member of WG21. Crowl asked how to get into that directory, Sutter recommended talking to Hedquist about anything related to the US NB. Voutilainen concurred Wong's view. Sutter said we'll still have PL22.16 and WG21 meetings at the same time, but we don't need to have PL22.16 votes separately, unless there's some PL22.16-specific business ongoing.
Simonsen said this shouldn't practically change anything, and suggested that the convener is free to ask NB individuals for guidance about what the NB is likely to think.
Wong reported that his constituents are worried about the possibility of being able to load up the room with a large number of people voting for a certain matter. Sutter said consensus has always been taken into account in addition to numerical count, and that NB input will still be queried for. Carruth said he doesn't think a single company hypothetically sending a lot of non-regulars into a meeting still would be able to sway votes. Sutter said that the convener still determines consensus, and pointed out that WG21 tries to avoid sending out ballot documents that would have a large chance of failing.
Sutter summarized the new document rules, pointing out that LiveLink is mandatory and other repositories are banned. He said WG14 is trying to move its documents to LiveLink. He also said it might be possible to continue using open-std.org to just contain links to LiveLink. He indicated that new papers may end up on LiveLink after July. Goldthwaite said she thinks nobody has ever been able to make LiveLink work for BSI, and voiced concern about the password management, and about LiveLinks compatibility with client platforms.
Plum pointed out that researchers and article authors want to provide permanent links to the papers without requiring registration and having links that work quickly and are stable. Voutilainen pointed out that the idea of moving existing papers has potential copyright issues. Sutter asked whether that applies to new papers, and Voutilainen said it's possible. Simonsen thought that since authors hold the copyright to their works, ISO can't dictate where the documents can or cannot be stored and/or published.
Sutter said that there are possibilities to comply with the LiveLink requirements without disturbing the way we are used to working. Crowl asked whether ISO will commit to universal accessibility, and Clow emphasized the need to access the documents without authentication. Sutter said he doesn't have an answer to that. Goldthwaite expressed concern about document formats that are accepted on LiveLink, Sutter said we have never submitted eg. word docs and shouldn't need to do so in the future either.
Meredith asked about the issue lists, and Sutter explained the directives apply only N-papers. van Winkel asked about the wikis, and Sutter said that is outside the purview of ISO. Simonsen asked whether this is going to be discussed in Rapperswil, and Sutter answered it's very likely but he hopes that it will be in a limited fashion, since the topic is mostly administrative.
WG14 is working on TS 18661 (IEEE floating point), their last meeting was in Parma, Italy, John Benito is stepping down as the convener, next meeting in St. Louis, new convener (David Keaton) starting there. WG14 is trying to comply with the new JTC1 document requirements.
The DIS ballot for C++14 closes Aug 15-18.
One older paper, N3871, needs to be put on the agenda, others being taken care of.
Koenig wishes to retire from maintaining the current reflectors. There are also quite many problems with the existing mailing list software. Sutter said there's likely going to be discussion between google groups and mailman, and that there have been concerns about long-term availability of google groups. Carruth assured that google groups are not going to go away. Carruth also said that there has been work on fixing the problems with emails being dropped or marked spam, and that both google groups and mailman have the sufficient technology to avoid these problems, and said that both of these options work fine, but other options are likely going to be insufficiently good.
Vollman pointed out that google groups links in emails require javascript. Sutter said that the decision needs to be done administratively in a small group with input from the larger group. Clamage said that we need to pay particular attention about solutions that meet strong opposition. Sutter suggested having some admin reflector discussions about the migration possibilities and options. Miller said it would be useful to collect feedback from people who have experience in mailman and google groups.
Carruth voiced concern about the scheduling of the subgroup sessions and study group sessions during the meeting week, and asked what the proper venue and timing for discussing that would be. Sutter suggested taking that discussion offline over food and beverages.
Will be in the post-meeting mailing.
Will be reviewed at the end of the Rapperswil meeting.
Will be discussed at the end of the Rapperswil meeting.
Sutter ended the meeting 10:15 Pacific Time.