C++ Logo

SG10

Advanced search

Subject: [SG10] Meeting Monday 03-17: minutes
From: Nelson, Clark (clark.nelson_at_[hidden])
Date: 2014-03-17 13:57:15


SG10 had its first WebEx-hosted meeting on March 17.

Attendees:
Clark Nelson (chair)
Richard Smith
Walter Brown
John Spicer
Jens Maurer
Tom Plum

We reviewed the proposed revision of SD-6 that Clark had posted to the wiki:

http://wiki.edg.com/twiki/pub/Wg21issaquah/SG10/sd-6.htm

The review consisted mainly of considering the highlighted entries in the
C++14 table. Some of the entries were accepted without discussion.

The discussion of digit separators concluded that a macro indicating their
availability may be useful despite the problems. There was no (serious)
suggestion for a different macro name.

During the discussion of the deprecated attribute, some sentiment was
expressed for a different mechanism for testing for the availability of
attributes, including non-standard attributes, along the lines of clang's
__has_attribute feature. Richard volunteered to post a description for
consideration and discussion on the reflector. (Until that discussion
happens, I am considering the entry for the deprecated attribute to be
unresolved.)
AI Richard

Decisions about exactly what to do about features that will be specified in
some Technical Specification are a little premature yet. Even if each TS
specifies its own feature-testing method, we may want SD-6 to contain
summaries. We will have more information after the Rapperswil meeting.
However, it was decided that the entries for the features removed from C++14
should at least be moved to a separate table in the C++14 section.
AI Clark: create the new table

It was noticed that the macro for complex UDLs should be defined in the
<complex> header, not the <chrono> header.
AI Clark: fix it

The situation with respect to shared locking is confused. The <shared_mutex>
header was added by N3659, not N3891. There is a question whether that paper
also changed functionality in the <mutex> header.
AI Clark: try to figure out what gives

At a higher level, there was some discussion of updating SD-6, including a
question about whether that would require a vote of WG21. The consensus was
that a WG21 vote is not necessary as long as all changes are consistent with
the approach approved in Chicago. (Adding __has_attribute could be
considered a departure from that approach.) It was also agreed that, under
the circumstances, more frequent updates would make the most sense.
AI Clark: confirm that Herb agrees with SG10 about SD-6

There was also discussion of the benefits of WebEx meeting technology. It
provides for video sharing, and dial-in numbers in more countries, over
Intel's audio-only bridges, but the audio quality tends to be not quite as
good. Overall, WebEx was preferred by/for SG10.

--
Clark Nelson            Vice chair, PL22.16 (ANSI C++ standard committee)
Intel Corporation       Chair, SG10 (C++ SG for feature-testing)
clark.nelson_at_[hidden]  Chair, CPLEX (C SG for parallel language extensions)

SG10 list run by herb.sutter at gmail.com