[cfp-interest 3732] JTC1/SC22/WG14/CFP 2025/11/12 Meeting notes

Tue Nov 18 17:28:55 PST 2025

2025/10/29: 8:00 AM PDT/: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, David Hough,
Joshua Cranmer, Fred Tydeman, Tue Ly

*Updated agenda plus new items*

https://wiki.edg.com/pub/CFP/WebHome/CFP%20meeting%20agenda-20251112-agenda-update.pdf
–
These minutes should be read alongside the agenda, with its many reference
links.

    Jim: Has an added item for last 15 minutes.

*Previous meeting notes*

   https://wiki.edg.com/pub/CFP/WebHome/n3735.pdf 2025/10/29 Meeting notes

*Study group logistics*

    Next meeting: 7 January 2026, 8:00 AM PST/4:00 PM UTC

    ISO Zoom teleconference

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

*C documents*

    The latest C2Y draft is N3685 Sept. 2025
https://www.open-std.org/jtc1/sc22/wg14/www/docs/n3685.pdf

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

*IEEE 754 liaison*

    Damian: Still discussing language of arithmetic and interchange
formats. No offline chat about complex and imaginary yet.

*C++ liaison*

    Joshua:Meetings happened last week. A question is coming to CFP on use
of cr_ prefix for C++ functions returning correctly rounded results. The
committee prefers adopting C type rules for std::complex heterogeneous
operations. Resolving fast-math is hard; consensus was to focus on
specifying floating types that would opt-out of fast-math rather than
trying to specify anything about fast-math. The committee is researching
new floating types (and expression evaluation) to guarantee reproducible
results across diverse platforms.

*WG14 update*

    Rajan: Next meetings 2-6 Feb 2026.

*TS-4 and TS-5 revisions*

    None.

*News*

    David: 754-1984 40th anniversary meeting for what was called KCS at the
(distant) time happened without any of Kahan, Coonen, or Stone able to
attend. Prof. dialed in to the meeting, with comments on the standard and
his trusty students who contributed over the years. Coonen mailed in his
intended historical remarks.

    Jerome: Videos of the Intel 8087 Milestone plaque unveiling in Haifa,
Israel have become available. I will coordinate with Leonard Tsai about
sending them to the 754 group, and then the CFP group.

*Carryover action items from last meeting *

   Jerome: Follow up on cfp3618 to see whether snan issues apply to more
than signbit, and what to do about it.

        Done, discussion.

*Action items from last meeting *
    Jim: Amend the nextafterd proposal with language about the
breaking-implementations issue, share with CFP, and then submit if there's
no objection.
     Done

    Jerome & Jim: Submit proposal about error handling and errno.
    Done

    Damian: Update paper on Annex F, based on long (converging) discussion
of the several alternatives, plus the specific discussions of atan2, frexp,
ilogb, hypot, compoundn, etc.
    Done, discussion.

    Group: Review Damian's coming Annex F changes and study pow and others
which weren't covered in the meeting, for lack of time.
        Done

    Damian: Clean up 3.14 "correctly rounded result" and research any other
usage of that or similar language.
        Done, closed.

    Jim: Draft change to references to rounding precision mode. Circulate
to new ad hoc rounding precision mode subcommittee: David, Damian, Jerome,
Jim.
        Done, but too late for discussion so carry over. Action.

*Discussion of issues*
    sNaN issues, signbit
        Jerome: Reviews the report email, looking for cases like signbit
needed extra care. The handling of NaNs is covered in F.2.2 and F.3, but a
pass over the many NaN references in the draft suggests that Damian's added
language is more helpful than redundant. We will make the final
determination in Damian's proposal. The cases copysign and fabs seem
adequately covered.
        Jim: Further describes the balance between stating qualifiers like
"quiet and signaling HaNs" vs. forcing the reader to infer the behavior
from arguably subtle specification.This is input for Damian's item F.3. And
Jim notes that the Reader's Guide to signaling NaNs was helpful for
reviewing the paper.

    Conflicting definitions of double_t
        Jim: Joseph Meyers has brought up a subtle issue where different
sections of the standard suggest that double_t ought to be double in one
case long double in another. Given Joseph's care in these matters, this
merit s a close look. Action, with Jerome a belated volunteer.

    Sign of zero from fromfp functions
        Jim: Joseph Meyers writes that the sign of the zero result is not
clear. When the functions were changed to have a floating-point result, the
sign of zero results naturally became an issue.
        Damian & Jerome: Action.

    "result" vs. "result value" vs. "return value"
        Jim: This and the next two items involve direct contact with the
editor. We will consider them as taken together in a note to JeanHeyd. Tney
are too inconsequential to warrant full proposals. First, the expression
"function value" is not best, leading to the three choices in the title of
this item. "Result" is deemed to be the best choice.
        Damian: Some of the instances of "function value" come up in my
changes, where the fix is part of the larger change.
        Rajan: Takes action to report to JeanHeyd, including the next two
items.

    SNAN macro location change
        Jim: Reminds group of moving the macro from math.h to float.h in a
prior draft.
        Fred: Glibc still has relevant macros in math.h, but appropriately
guarded by version macros. The float.h is the newer spelling.
        Jim: The move wasn't recorded in the Annex M changes. Append to
Rajan's action.

    strtodN issue
        Jim: Fred raised the issue of the discussion of conversion in the
standard without mention of the type converted to. "Convert to a number"
should be "convert to a number of type..." or something similar.
        Rajan: Suggests an explanatory sentence because the change is more
elaborate. Appends to action.
        Rajan & Jerome: Suggest changing the wording of cfp3688 slightly to
"...of the return type of the function..."

    Rounding precision mode, Annex F floating-point environment
        Jim: Deferred to next meeting.

    Annex F special cases
        Damian: Reviews many changes, picking up where last meeting ended.
Highlightd that the NOTE has been split into NOTE 1 and NOTE 2.
        Jim: Need to change the sign character in "-y" in first item of
atan2 and atan2pi to a hyphen.
       Jim: Use "is equivalent to" for pow(+/-INF, n) rather than defining
it in terms of a C expression, which is not the style of the draft.
        Jim: Will need to look at F.10.79, the trunc functions, with regard
to rounding. There is some subtlety.
        Fred: In canonicalize, use "is" not "must".
        Jim: A nice goal is to promote this to a proposal after the next
meeting in January.
        Damian: Action.

    Rounding modes, terminology
        Jim: Damian has pointed out many inconsistencies in language of
"rounding direction" and "rounding direction mode" and even "rounding mode".
        Damian: Elaborates and points out that Fred touched on this 20+
years ago. It's an issue.
        Jim: Presents language based on 60559. Takes action to initiate
online discussion to tighten up this usage.

    Jim's bonus added agenda item on CFP futures
        Jim: What is CFP's purpose.
        Rajan: WG14 consensus is for CFP to handle everything related to
floating point. Includes maintenance and editorial revision.
        Jim: It's related to what is the purpose of the C standard. (a)
Reference for implementers and programmers. (b) Seeks to have well-defined
terms and is readable by readers with basic English knowledge. These are
two extremes.
        Rajan: Agrees with (a). Says that WG14 does not adhere to (b).
         Jim: Recalls P. J. Plauger's famous test when changes were
suggested: "Whose ox is gored?"
        Jerome: Another prospect facing CFP is expansion into entire
technologies like the expansive correctly-rounded library function project
and bit-reproducible results across divers hardware and OS platforms.
        Jim: Assigns a group action to ponder CFP in 2026.

*Other issues*
    Ly (submitted in Chat sidebar): We wish to expand complex.h to support
all of the functions in math.h.

    Jerome (added in proof): We agree that the hyphen character represents
the minus sign in fixed-width code font. There is some question about the
proper character in ordinary text.

*Adjournment*

    10:05 AM PDT

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

*Action items to be carried over*

    None

*New action items*

      Group: Review Jim's latest doc (to be released) on rounding precision
mode and be prepared to review.

    Jim & Jerome: Investigate Joseph Meyers's issue of the conflicting
requirements for double_t to be either double or long double.

    Damian & Jerome: Look at +/-0 in fromfp, and also the case of a value
rounding to zero.

    Jim & Rajan: Send email to JeanHeyd about the editorial change of
"function value to "result". Include mention of the move of the SNAN macro,
which was not recorded in Annex M. Third, recommend the wording in cfp3688
wording with "...of the return type..."

    Fred: Send Rajan wording for the previous action, re. the moving of the
SNAN macros.

    Damian: Remember to update the date in new drafts of the working
documents. (Public reminder in the holiday spirit.)

    Jim & group: Review trunc and canonicalize in Damian's Annex F update
to come.

    Group: Review the rounding direction language, with Jim to provide an
email to kick things off.

    Group: Think about where CFP is headed.

*Discussion issues to be carried over*
    Rounding precision mode, Annex F floating-point environment

    Editorial issues F.2.2

    Editorial issues F.3, etc.

    Editorial issues F.10.1

    More Annex F special cases

*Signoff*
    Respectfully submitted. Great to see all the usual suspects gathered
for the final meeting of 2025.

-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/20251118/112d9c18/attachment-0001.htm>