[cfp-interest 3464] JTC1/SC22/WG14/CFP 2025/05/14 Meeting minutes

Tue May 20 09:43:25 PDT 2025

2025/05/14: 8:00 AM PDT/3:00 PM UTC
[Please submit proposed changes to these minutes to Jerome or to the group.
Revision changes appear at the bottom.]

*Attendees*

    Jim Thomas, Rajan Bhakta, Jerome Coonen, Damian McGuckin, Guy Davidson,
Tue Ly, David Hough

*Updated agenda plus new items*

   https://wiki.edg.com/pub/CFP/WebHome/n3553.pdf – These minutes should be
read alongside the agenda, with its many reference links.

*Previous meeting notes*

    https://wiki.edg.com/pub/CFP/WebHome/n3552.pdf

*Study group logistics*

    Next meeting: 11 June 2025, 8:00 AM PDT/3:00 PM UTC

    ISO Zoom teleconference

    Please notify the group if this time slot does not work.

*C documents*

    The latest C2Y draft is N3550 May 2025
https://www.open-std.org/jtc1/sc22/wg14/www/docs/n3550.pdf

    C23 has been published ISO/IEC 9899, available for purchase.
https://www.iso.org/standard/82075.html

*IEEE 754 liaison*

    Damian: No meeting recently due to Arith2025, where Jim Demmel won best
paper.

*C++ liaison*

    Jim: Guy Davidson sent a note about reproducible floating point, to be
discussed below.

*WG14 update*

    None

*TS-4 and TS-5 revisions*

    Jim: MInor issue, to be discussed below.

*Carry-over action items from last meeting *

    Damian: Update the proposal re. the wording of cimag() and creal().
(This is part 3 of Damian's larger effort.)
        Damian: Will post draft on wiki. Carry over.

    Ly: Send email to CFP about the next____() issues pertaining to the
several functions inspired by nextafter().
        Ly is sending during the meeting, so the item will carry over.

*Action items from last meeting *

    Jim: Submit updated proposal re. frexp clarification.

        Done n3535.

    Rajan: Propagate 3413 with wording just approved up to WG14, for
expression evaluation at translation time.
        Done email link in agenda.

    Damian: Submit the cproj proposal with minor amendments discussed. Jim
to get a document number and help submit.
        Done, will send to WG14.

    Jim: Submit email about TS response re. domain and range errors. Rajan
to work with issue tracker when possible.
        Email sent. Carry over to monitor tracker.

    Jim & Damian: Jim to send to CFP an email with suggested wording
changes for complex.h. Damian to update the proposal based on these and
other suggestions. Jim to get a document number for Damian, then submit the
updated proposal.
        Done, Damian will post n3537 draft for email approval.

    Jerome: Write a proposal for powr() addressing Joseph's suggestion.
        Done, 3452.

    Damian: Will check the powr() language against his ongoing C2Y changes.
        Done and closed.

    Jim: Keep an agenda item re. the removal of “(also known as a
singularity or infinitary)” in post-N3467 draft of C2Y.
        Jim: Fixed in N3550 draft. Closed.

*Discussion of issues:*

    Reproducible floating point

        Guy: Summarizes key issues in the letter linked to in the agenda.

        Rajan: Suggests looking at Clause 9 in ISO/IEC TS 18661-5 for
reproducible results.

        Jerome: Terminology complicates this discussion. (Please read as
facts not criticism, in CFP tradition.) Guy refers to arithmetic operations
and square root as "functions" not "operations". They are correctly rounded
in 754, vs. the transcendental functions, whose correctly rounded variants
are a different topic and not Guy's focus. "Reproducible" in the last
decades has meant bit-for-bit (see Java, e.g.), while Guy is trying to
defend against compiler abuse such as ignoring parentheses in "(a+b)+c".

        Ly: Introducing a new type (that presumably would require compilers
not to commit dangerous optimizations) will be a pain, esp. in C++.
Usually, it's a block of computation, so a pragma can localize the control.
Maybe there could be some magic to disable certain optimizations.

        Jim: C doesn't allow value-changing optimizations. Compile flags
might relax the rules, but the behavior Guy is reporting in non-conformant.

        Jerome: Is there a catalog of the different kinds of problems?

        Guy: Not yet. We have seen associative operations reordered. The
"as if rule" governing code alteration may differ or not be honored in our
contexts. Will send an updated version of the working paper for future
discussion.

    [[reproducible]] example in 6.7.13.8 ignores exceptions
        Jim: Side-effects like flags can be lost if a function is called
twice with identical arguments, for example. If the flags are cleared
between the calls and the second call is elided, then the presumed flags
from the second call will be lost. Sent an example to JeanHeyd with a
version of the function including save/restore of the state.
        Rajan: Suggests adding a comment to indicate why the flag
operations are there. Action.
        Jim: Suggests changing the content example to use sqrt(fabs(x))
which does not overflow or underflow. Will draft a response to Jean-Heyd.
Action.

    Domain and range errors in TS-4
        Jim: Wrote letter with a suggested fix, as linked in the agenda.
Rajanwill track WG14 response. Action.

    powr(x,y) with x < 0
        Jerome: Reviews simple "including NaN" change.
        Jim: Suggests updating to latest C draft, checking all section
numbers, and getting a document number. Action.

    cproj proposal
        Discussed above. Action.

    complex.h intro proposal
       Discussed above. Action.

    Return value vs return type
        Jim: Misuse of return value" occurs in other places. A bigger issue
is an action to find all instances of "return value" when meaning "return
type".
        Jerome: Offers to research and summarize. Action.

    Range error wording for atan2 and atan2pi
        Jim: Javier has proposed some clarification of the wording (with no
technical change).
        Rajan: Suggests letting Javier take this, if he wants. Javier might
choose to leave this to CFP. Action.

    Semantic rules for constant expression evaluation
        Jim: Suggests adding the comment, "unless otherwise specified", to
highlight some subtlety. Action.

    Editorial suggestions incl wording for cimag and creal
        Discussed earlier in Carry-Over Action Items.

    Imaginary I macro and i suffix issues, N3390 update
        Rajan: No word yet, but expect paper to come. Carry-over discussion
item.

    next____() issues
        Discussed under Carry-Over Action Items.

    More complex functions, e.g. ccbrt, csinpi
        Ly: Looking at expanding complex.h to the extent of math.h. Will
send email. Action.

    CFP list of post C23 issues
https://wiki.edg.com/pub/CFP/WebHome/c26g.htm

        Jim: This list descends from the basis of CFP's original work.
Interested in how to prioritize outstanding issues. WG14 is top of the list.

*Other issues:*
    None

*Adjournment*

    10:05 AM PDT

-----------------------------------------------

*Action items to be carried over:*

    Damian: Update the proposal re. the wording of cimag() and creal().

    Ly: Send email to CFP about the next____() issues pertaining to the
several functions inspired by nextafter().

    Rajan: Monitor the new WG14 issue tracker re. TS 18661-4, the domain
and range error issue.

*New action items:*

    Jim: Draft a note about the [[reproducible]] issue.

    Jerome & Jim: Get a document # for the powr proposal , 3550, check the
minor updates, and submit to WG14.

    Damian & Jim: Get a document # and submit the cproj proposal with minor
updates to WG14.

    Jim: Review the complex.h proposal from Damian to confirm that the
changes listed in CFP 3448 are correct, and there if are no other
significant changes then submit it to WG14. Otherwise, CFP will review it
again next month.

    Jerome: Search C2Y for instances of "return value" vs. "return type"
and summarize the usage across the whole document.

    Rajan: Send note to WG14 supporting Javier's changes re. atan2 and
atan2pi range errors, and encouraging Javier to proceed with his proposal.

    Jim: Write up a suggestion re. semantic rules for constant evaluation.

    Ly: Send email on expanded complex functions.

    Jim & Group: Write a note about the priorities list while group ponders
list.

*Discussion issues to be carried over:*

    Imaginary I macro and i suffix issues, N3390 update

Respectfully submitted.

-Jerome Coonen
 650.996.4738
 jcoonen at gmail.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.oakapple.net/pipermail/cfp-interest/attachments/20250520/119af0b7/attachment.htm>