Title: Standards Report

Author: Martin Moene

Date: 01 November 2014 15:33:36 +00:00 or Sat, 01 November 2014 15:33:36 +00:00

Summary: Mark Radford brings the latest news from C++ Standardisation.

Body: 

Hello and welcome to my latest standards report.

Once again the timing of this report is a bit off. Why? Because the big news should be that C++14 has become an international standard. However, because of the timing, the chances are that everyone reading this already knows about it (CVu has a production deadline approximately a month before it appears in print, therefore I submitted my previous report for publication just before C++14’s ratification). At the end of my last report, I said that C++14 was at its Draft International Standard (DIS) stage and, if none of the national bodies submitted a ‘no’ vote, then it could become an international standard. That is what happened: by the end of the DIS period (August 15th) all the member countries’ national bodies had voted ‘yes’, and the DIS became the ISO standard for C++14.

The above could happen owing to a recent change in ISO procedures aimed at helping to speed up the release of standards. Previously, if any national bodies issued comments with their vote on the DIS (even if it was a ‘yes’ vote) then there would be a Final Draft International Standard (FDIS), and another round of voting. With the recent change, if the ‘yes’ vote is unanimous – even if there are comments, which there can be, even with a ‘yes’ vote – an FDIS will be issued only if the standards committee explicitly requests it.

Is there anyone out there with strong Ruby knowledge? IST/5 – that’s the BSI committee handling programming languages – is interested in enrolling a Ruby representative. I don’t have any more details: the obvious assumption is that this would/could lead to the formation of a BSI Ruby Panel (just as there is a BSI C++ Panel). In any case, if there is anyone out there able to (and interested in) represent Ruby on IST/5, feel free to contact me in the first instance and I’ll put you in touch with the right person.

I last reported on C standardisation a couple of reports ago, so here’s what’s been happening since. Part 1 of the floating point TS is now published (last time it had been approved but was awaiting publication). Part 2 (which, last time, had recently been moved to the DTS stage) has now been approved for publication, but is not yet published (but that’s just a matter of time). Parts 3 and 4 have gone through their PDTS stage: the next WG14 (ISO C Committee) will consider responses to ballot comments on the PDTS. There is a part 5 to come, but as yet no draft is available. There is, however, a public draft of the document ‘Programming languages – C – Extensions for parallel programming’ (N1862) available, which can be found at: http://www.open-std.org/JTC1/SC22/WG14/www/docs/n1862.pdf. I assume this is to be a TS (it looks like that sort of document to me), but it doesn’t say so on the document and I don’t, as yet, have any clarification.

The next ISO C++ meeting will be held in Urbana-Champaign, IL, USA, 3rd–8th November (which means by the time this report is published the meeting will have already happened). I see that the mailing for this meeting is not yet published (and probably won’t be before I submit this report for publication), so I won’t be talking about any of the papers included in it. There was a face to face meeting of SG1 (the study group for concurrency and parallelism) in early September, and I can report briefly on day one of that meeting (sorry but I don’t have any information about day two). More of that below, but first let me mention that the transactional memory extensions TS passed the ballot needed for the new work item to be official. Therefore we now have: ‘TS 19841: C++ Extensions for Transactional Memory’.

Moving on to the SG1 meeting, in my previous report I pointed out that the ‘Executers and Schedulers’ section had been removed from the draft concurrency TS (N4107). Until its removal, that section had come from the proposal ‘Executors and schedulers, revision 3’ (N3785). It was removed to allow for the consideration of the ‘competing’ proposal ‘Executors and Asynchronous Operations’ (N4046) by Christopher Kohlhoff, the latter proposal having received a very positive reception at the last ISO meeting (Rapperswil). That’s where things had got to before the SG1 meeting. When the meeting took place a draft follow-up to N4046 was available, containing actual wording for the TS (N4046 did not contain wording, confining itself to a description of the proposal, to check the level of interest). Also available was a draft revision of an N3785 follow-up, updated in response to N4046.

So far, so good, but then things started to go wrong. Originally there was going to be no facility to attend the meeting remotely. Although this facility was provided at the last minute, it was too late for Christopher Kohlhoff (who was not able to attend in person) to be able to attend remotely. On the other hand Chris Mysen (N3785 co-author) attended in person, and therefore was able to represent that proposal and its draft follow-up.

The discussions ended in a straw poll, the result of which (fourteen in favour, two against) suggested that Chris Mysen’s proposal should serve as the starting point, and what parts of Chris Kohlhoff’s proposal cannot be implemented in terms of it should be researched. The results are to be discussed at the forthcoming Urbana-Champaign ISO meeting. I don’t understand this as it seems to be starting in the wrong place, because Chris Kohlhoff’s proposal has a far better track record of standing up to scrutiny than Chris Mysen’s.

Before wrapping up I have some acknowledgements. Thanks to Roger Orr for drawing my attention to the news about the transactional memory TS, and for letting me know about the change in ISO procedure that allowed C++14 to become an international standard without the need for an FDIS. Thanks to Jamie Allsop for his excellent report (to the BSI C++ Panel) on day one of the SG1 meeting (which he was able to attend remotely); I have drawn heavily on Jamie’s report for my reporting on this meeting. Finally thanks to BSI C Panel convenor Joseph Myers for updating me on C standardisation progress.

Notes: 

More fields may be available via dynamicdata ..