JTC1/SC22
N2306
Date: Wed, 9 Oct 1996 09:55:53 -0400 (EDT)
From: "william c. rinehuls" <rinehuls@access.digex.net>
To: sc22docs@dkuug.dk
Subject: Document SC22 N2306 - WG21 Minutes
_________________________beginning of title page ________________________
ISO/IEC JTC 1/SC22
Programming languages, their environments and system software interfaces
Secretariat: U.S.A. (ANSI)
ISO/IEC JTC 1/SC22
N2306
October 1996
TITLE: Minutes of SC22/WG21 (C++) Meeting on July 7-12,
1996 in Stockholm, Sweden
SOURCE: Secretariat, ISO/IEC JTC 1/SC22
WORK ITEM: N/A
STATUS: N/A
CROSS REFERENCE: N/A
DOCUMENT TYPE: WG21 Meeting Minutes
ACTION: To SC22 Member Bodies for information.
Address reply to:
ISO/IEC JTC 1/SC22 Secretariat
William C. Rinehuls
8457 Rushing Creek Court
Springfield, VA 22153 USA
Tel: +1 (703) 912-9680
Fax: +1 (703) 912-2973
email: rinehuls@access.digex.net
_______________end of title page; beginning of text ______________________
Minutes
ISO/IEC JTC1/SC22/WG21 Meeting No. 16
7 - 12 July 1996
Sheraton Hotel
Stockholm, Sweden
1 Welcome from convener and host
Harbison convened the meeting at 18:05 (MESZ) on Sunday, 7 July 1996.
Saks was the secretary.
1.1 Introduce technical experts
The attendees introduced themselves. The attendance list appears as
Appendix A.
1.2 Adopt agenda
Harbison presented the proposed agenda SD-0 (revised 28 June 1996), and
recommended adding the following item:
2.4 Project Schedule
WG21 accepted the agenda with this changes.
1.3 Select drafting committee
Saks explained that the drafting committee normally meets on Wednesday
evening to draft the formal motions. Unfortunately, our host scheduled
a reception for Wednesday evening, so we need to schedule the drafting
work around it.
Koenig suggested that the drafting work should be part of the scheduled
agenda, possibly during the daytime on Thursday. Saks agreed that
drafting should be on the agenda, but doing the working in the evening
is a more effective use of committee time.
After some discussion, the WG agreed to do as much drafting work as
possible on Monday and Tuesday evening. Saks said we should expect some
of the drafted motions to change as a result of Wednesday's discussion,
but we should still try to get something in writing for each motion by
Tuesday night.
As usual, Saks will coordinate the drafting committee. Corfield,
Hartinger, Rumsby and Unruh offered to participate.
1.4 Approve minutes from previous meeting
Saks submitted N0879 = 96-0061 for approval as the minutes of the
previous meeting. WG21 accepted the minutes without change.
1.5 Review action items from previous meeting
None.
1.6 Recognize documents
Harbison introduced SC24 N 1577 (see item 2.2).
2 New business
2.1 Convener Change
Harbison said he will not serve another three-year term as WG21 con-
vener. The US national body holds the WG21 convenership, and will
likely nominate Plum to serve as convener. (A ballot to recommend Plum
is still in progress in the US TAG.) If all goes as planned, SC22 will
approve Plum as WG21 convener at the SC22 plenary in September, 1996.
Harbison said Plum would then resign as US IR (international Repre-
sentative).
2.2 Liaison reports
Soop presented a liaison statement from SC24/WG6 (SC24 N1577). WG6 is
developing a standard for object-oriented multimedia functionality
(PREMO). They have identified extensions to the C++ language that they
believe they need to create a PREMO binding for C++, and that may be of
general use. Specifically, WG6 would like WG21 to consider extending
the C++ language to support concurrent processing and synchronization.
Some WG21 members asked why WG6 thought they needed language extensions
(as opposed to library facilities) to support their needs. Soop recom-
mended that WG21 members contact Ivan Herman from NNI (Netherlands) and
Jim Van Loo from ANSI (USA) for details.
Bruck volunteered to look into the issue on behalf of WG21 in coopera-
tion with Soop.
Stroustrup mentioned that there's an 800-page report from MIT Press
(edited by Wilson) which describes 18 different approaches to concur-
rency in C++.
Glassborow thought we should stay focused on getting the draft out, and
consider this request only after we have completed the standard. Other
WG21 members agreed. Bruck said that was desirable.
Harbison agreed to explain WG6's request to X3J16, and draft a response
to WG6 along the lines that Glassborow suggested.
Soop expressed concern that different people are implementing multi-
threading extensions in different ways. He would like WG21 to say it
will address this eventually. Stroustrup had doubts that everyone could
agree on a single standard approach to concurrency in C++.
2.3 CD ballot resolution status
Harbison said he's relying on each NB (national body) delegation to tell
him if they are satisfied with WG21+X3J16's handling of their comments
on the last CD (committee draft) ballot. He invited comments from each
delegation.
Kamimura said the progress on extended characters is satisfactory, as is
the progress on traits and strings, and on template compilation.
Lajoie said Canada wants to advance the CD, more than it is concerned
about any particular resolution to template compilation issue.
Hartinger reported that Germany is still concerned about the template
compilation model. He said it's not clear if the new proposal (from
SGI) can be ready in time to submit the draft for another CD ballot this
week. If it is not ready this week, they would prefer to return to the
Santa Cruz proposal (to allow inclusion only). Germany is also
concerned that iostreams is not ready for CD (it has too many open
issues). Hartinger added that some of Germany's smaller issues (mostly
in the library) did not yet get a response from WG21+X3J16.
Plum explained that he, Schwarz, Clamage, Myers, and le Mouel met in San
Jose, CA several weeks ago to handle most of the 70 open iostreams
issues. They drafted proposed resolutions which they will present to
the Library WG this week. Harbison asked Hartinger to review Germany's
ballot comments to see which have not been addressed.
Bruck said Sweden has two main concerns: 1) keeping the project on
schedule, and 2) the template compilation model. He said the SGI
proposal appears to be a major step toward resolving (2). Some of
Sweden's other issues have been handled, some will "drop on floor", and
at least one (division of negative numbers) has been handled by the C
committee (SC22/WG14).
Rumsby said most of the UK's issues have been handled. Others that
haven't are just editorial nits. Lajoie said she thought all the UK's
issues had been handled. Rumsby said he wasn't sure, but in any event,
none of the issues would prevent the UK from voting to advance the draft
to the next CD ballot.
Harbison emphasized that holding a third CD ballot would be a BAD THING.
This second CD should have essentially the form and content of the
eventual standard. We should believe that any comments from NBs could
likely be handled by localized edits to the draft.
Rumsby said the UK is concerned that incorporating the SGI proposal on
template compilation would be too big a change for them to vote for CD
this week. However, approving the Spicer/Ball proposal (tabled from the
last meeting) would not.
Glassborow said he did not want to vote for another CD until he could
see the actual text of the draft. He preferred to vote at the start of
the next meeting (in November).
Koenig supported Glassborow's position. He further suggested that we
begin the next meeting with a formal motion to approve the CD (possibly
with amendments to correct errors.) He added that this does not slip
the project schedule by a full meeting (four months), because WG21 could
submit the draft for CD immediately after that meeting. There would not
be the usual six-week delay to complete the editing after the meeting.
Rumsby said he thought Koenig's suggestion is what the UK had in mind.
Harbison said that Koenig's proposal still slips the schedule by a full
meeting unless we change our meeting dates a little.
Plum said he personally likes Glassborow's proposal, he's concerned that
the world will just know that the C++ standard has slipped its schedule
again.
Koenig said he may be relocating his office in next few months and can't
commit to any schedule for delivering an updated draft before the
November meeting.
Stroustrup asked that WG21 recommend that the technical work on the
template compilation model be aimed at achieving consensus based on the
SGI proposal.
Plum wanted to make it clear that all we are doing is delaying the
editing meeting until the start of the November meeting. We don't want
to delay the solution of any problems until next meeting. Others agreed
that WG21 would like all substantive changes to C++ to be made by the
end of this week.
3.0 Other business
None.
4 Closing
4.1 Assign permanent document numbers
See SD-1 for document numbers.
4.2 Review action items and issues
None.
4.3 Recess
WG21 recessed at 20:55 on Sunday and reconvened in joint session with X3J16.
See the corresponding WG21+X3J16 meeting minutes (N0880 = 96-0062).
Appendix A - Attendance
Name Affiliation; (*) = Head of Delegation
Lajoie, Josee Canada (*)
Harbison, Sam Convener
Stroustrup, Bjarne Courtesy
Hartinger, Roland Germany (*)
Unruh, Erwin Germany
Kamimura, Tsutomu Japan (*)
Koshida, Ichiro Japan
Umekawa, Ryuichi Japan
Bruck, Dag Sweden (*)
Jonsson, Fredrik Sweden
Soop, Karl Sweden
Corfield, Sean UK
Glassborow, Francis UK
Rumsby, Steve UK (*)
Southworth, Mark UK
Clamage, Steve USA / X3J16 Chair
Koenig, Andrew USA / Project Editor
Plum, Thomas USA (*)
Saks, Dan USA / Secretary
Plauger, P. J. WG14 Convener
_______________________end of document SC22 N2306 ______________________