[Cfp-interest 2131] Re: Preliminary results of the WG14 (C Standards meeting) papers of interest to CFP

Jim Thomas jaswthomas at sbcglobal.net
Sat Sep 4 12:29:39 PDT 2021


Rajan and Fred, thanks for representing CFP, and the good progress.

Rajan, thanks for the detailed report.

A few comments below …

- Jim Thomas

> On Sep 2, 2021, at 8:27 AM, Rajan Bhakta <rbhakta at us.ibm.com> wrote:
> 
> Summary: We had almost everything pass and make it into C23. The one exception is the numerically equal paper (N1716) where our second change needs to be reworded and the fenv paper (N2748) where we need to reword the first change (as recommended practice for example).
> We also did not cover N2746 (overflow/underflow redefinition) since we have updated text for it that did not make it in in time and N2747 due to both time constraints and lack of clarity on whether we needed CFP2090 in or if that changed this paper.
> 
> Raw details below:
> 
> 5.12 Tydeman, Overlooked SNAN wording changes [N 2710]
>   Straw poll: Does WG14 want N2710 in C23 (with LDL -> LDBL)?
>     16/0/2. Goes into C23.
> 
> 5.13 Tydeman, fmin, fmax [N 2711]
>   Straw poll: Does WG14 want N2711 in C23?
>     11/0/6. Goes into C23.
> 
> 5.16 Tydeman, hypot() [N 2714]
>   Fred: Typo: Should be "returns *a* NaN" for consistency with the rest of the standard.
> 
>   Straw poll: Put N2714 into C23?
>     17/0/2. Goes into C23
> 
> 5.17 Tydeman, cr_ prefix [N 2715]
>   Straw poll: Put N2715 be put into C23 with "potentially" being kept in (not removed)?
>     15/0/5. Goes into C23.
> 
> 5.18 Tydeman, Numerically equal [N 2716]
>   Jens: I think the the two first cases would be clearer if they say "the results compare equal".
>   Alex: The first one is good, since it is number based. But the later ones, do need a modifier since it reads more that it applies to numbers.
>   Rajan: This is forward progress. May not be perfect. We can have Jens and Alex write papers or have it go back to CFP.
>   Roberto: Is numerically equal vs equal re +/-0 needed?
>   Freek: Is a NaN numerically equal to itself?
>   Fred: No.
>   Freek: This implies numerically equal makes a difference.
>   Joseph: IEEE has different levels of equal and equality. If we want to change this, we could align with them.
>   Rajan: CFP did consider the -ve zero case. We convinced ourselves that this change did not affect that.
>   Joseph: Perhaps we should have definitions for equal, equivalent. ISO has terms but not these ones. We could add something to clause 3.
>   Jens: I find the second case confusing. Having it compare equal is better.
>   Fred: We can take this back to CFP.
>   Roberto: These are the only two occurrences so I am OK with the changes except the 2nd one. I agree with Jens there.
> 
> 
>   Straw poll: Does WG14 want changes 1 and 3 in N1716 in C23?
>     13/0/5. Goes into C23.
>   Straw poll: Does WG14 want change 2 in N1716 in C23?
>     1/3/13. No consensus.
> 
We can change the 2nd one to say “compare equal”. Our to say “equal (i.e., compare equal)” to avoid suggesting that “equal" and "compare equal" are different..
>   Straw poll: Does WG14 want something along the lines of change 2 in N1716 in C23?
>     11/0/6. Consensus.
> 
> 5.19 Thomas, C23 proposal - range error definition [N 2745]
>   Straw poll: Put N2745 into C23 without the footnote?
>     12/0/7. Goes into C23.
> 
> 5.20 Thomas, C23 proposal - overflow and underflow definitions [N 2746]
>   New version based on CFP 2090.
>   Fred: We are withdrawing this paper for a replacement.
>   
> 7.1 Thomas, C23 proposal - Annex F overflow and underflow [N 2747]
>   Did not do this paper due to lack of clarity on changes from upcoming main standard updated paper on overflow/underflow from CFP and interaction with this paper.
> 
Above two awaiting document number for update to N 2746. I requested the number too late.
>   
> 7.2 Thomas, C23 proposal - effects of fenv exception functions [N 2748]
>   Keaton: "Should" should not be in a footnote. You could use "could". You could also make it normative.
>   Ballman: Should implies you should do it, Could is maybe it could happen.
>   Joseph: If we move it out of a footnote, it could be recommended practice.
>   Keaton: That does work.
>   
>   Straw poll: Put something like change 1 as recommended practice in N2748 into C23?
>     15/0/3. Clear direction.
> 
Hmm. Having recommended practice for implementation extensions seems odd. How about changing “should be honored” to “would be honored”? Or “is intended to be honored”?
>   Straw poll: Put N2748 changes 2 and 3 into C23?
>     16/0/2. Goes into C23.
> 
> 7.3 Thomas, C23 proposal - IEC 60559 binding [N 2749]
>   Straw poll: Does WG14 want N2749 in C23?
>     11/0/6. Goes into C23.
> 
> 7.4 Tydeman, static initialization of DFP zeros [N 2755]
>   Straw poll: Put N2755 into C23? (With typo corrections: implemen*t*ation)
>     14/1/3. Goes into C23.
> 
> Regards,
> 
> Rajan Bhakta
> z/OS XL C/C++ Compiler Technical Architect
> ISO C Standards Representative (Canada, USA), PL22.11 Chair
> C/C++ Compiler Development
> rbhakta at us.ibm.com
> 
> IBM
> 
> _______________________________________________
> Cfp-interest mailing list
> Cfp-interest at oakapple.net
> http://mailman.oakapple.net/mailman/listinfo/cfp-interest

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.oakapple.net/pipermail/cfp-interest/attachments/20210904/e534c844/attachment.htm>


More information about the Cfp-interest mailing list