[cfp-interest 3690] JTC1/SC22/WG14/CFP 2025/10/29 Meeting notes

Fri Oct 31 16:21:16 PDT 2025

2025/10/29: 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, David Hough,
Joshua Cranmer

*Updated agenda plus new items*

https://wiki.edg.com/pub/CFP/WebHome/CFP%20meeting%20agenda-20251029-update.pdf
– These minutes should be read alongside the agenda, with its many
reference links.

#2

*Previous meeting notes*

   https://wiki.edg.com/pub/CFP/WebHome/n3729.pdf 2025/10/15 Meeting notes

*Study group logistics*

    Next meeting: 12 November 2025, 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: Further discussion of the wording re. formats.

*C++ liaison*

     Joshua: Next meeting first week of 3 Nov. Two papers on floating point
topics: defining fastmath semantics is hard and reproducibility.

*WG14 update*

    Rajan: Next meetings 2-6 Feb 2026.

*TS-4 and TS-5 revisions*

    None.

*News*

    Jerome: 754-1984 40th anniversary meeting is hastily moved from Sat 1
Nov to Sun 2 Nov to avoid collision with Cal football game.

*Carryover action items from last meeting *

   Jim & Jerome: Look further into the language around when errors "occur"
in the sense of C2Y and how the setting of errno fits in. Draft a proposal,
if appropriate.

        Done

*Action items from last meeting *
    Rajan: Submit the proposal regarding translation time expression
evaluation.
        Done n3732

    Jim: Submit the proposal regarding output bounds for math functions.
        Done n3731

     Jim: Draft a proposal for the language of quantum exponent in
nextafterdN.
        Done

    Damian: Draft Annex F change proposal in two flavors, with modifiers
and with double-ended bounds.
        Done

    Jerome: Follow up on cfp3618 to see whether snan issues apply to more
than signbit, and what to do about it.
        Carry-over, after much offline discussion with Jim.

    Jim: Write a note about the locution of "result" vs. "result value" vs.
"return value" in some problematical functions where the language is not
best today.
         Done, in discussion.

    Jim: Add strtodN to list of low-priority issues.
        Done, in discussion.

*Discussion of issues*

    Preferred quantum exponent for nextafterdN

        Jim: Reviews the language emphasizing the numerical value (vs.
representation), in order to bring out the preferred quantum exponent.

        Rajan: As mentioned last time, this breaks breaks an implementation
that just returns y.

        Jim: Action to add mention of the breaking-implementations issue,
for purposes of WG14 clarification, and then submit if no objection.

    F.3 #7 quiet/signaling NaNs
        Jerome: Will follow up on cfp3618 to see whether signaling nan
issues apply to more than just signbit, and what to do about it. (NOTE: By
popular demand, will start using the "cfp" prefix on CFP-specific docs like
the email cfp3618 in this item. This removes ambiguity with WG14 docs
precede by n".)

    Overflow, non-default rounding, errno
        Jerome: Summarizes the changes to (a) emphasize the notion of
"error occurs"; (b) consolidate the specification allowing
implementation-dependent error reporting; (c) clarifies that the language
of over/underflow applies only when the exception is reported.
        Jim: Shows how footnote 275 about exceptions involving infinite
operands is not required in the specification.
        Jerome & Jim: Action to submit.

    Annex F changes discussion paper
        Damian: Walks through the options for pure inequalities vs. mixed
modifiers and inequalities
        Group: Much discussion of the issues. Prevailing sentiment was that
arguably verbose expressions like
            -inf <= x <= +inf
to express "all numbers, finite or infinite" is better than the more terse
expressions with cryptic absolute values. The rare cases of "all values,
even NaNs" necessarily require some verbiage.
        Rajan: WG14 sometimes gets big changes, so it won't be
problematical to get a possible update to Annex F with many update.
        Group: The issue of footnotes being specific to a single item leads
to the suggestion that we use a Note in the body text for matters that
pertain to multiple items.
        Damian: The cases atan2(y,0) and atan2(0,0) are problematical
because atan2 is described as computing "the arc tangent of y/x".  I've
added language such as "...and raises no "divide-by-zero" floating-point
exception to emphasize the case for atan2(y,0) for finite, nonzero y.
        Jim: The standard normally describes when errors DO occur, not when
they don't.
        Damian: In frexp, based on suggestion, propose removing an existing
line about raising not floating-point exceptions if the result is not a
signaling NaN.
        Jim That might be tied to double double. Best to leave it.
        Damian: The change to ilogb is just a reordering of the current
language.
        Group: Deemed good.
        Jim compoundn should not mention "integer" because that implied by
the argument n.
        Damian: The hypot(NaN, NaN) case is subsumed into the case
hypot(x,NaN) with |x| != inf.
        Damian: Actions.

    Rounding modes, terminology
        Damian: Suggest change across C2Y to be consistent about "rounding
direction mode" terminology, given that there is also a "rounding precision
mode".
         Rajan and group: Yes, good idea.
        Jim: Do we really need to define "result format", a phrase in the
current 3.14 (ed. note -- using the numbering in C2Y n3685)?
        Jim: After much back and forth, suggest changing the expression to
eliminate the usage "result format".
        Jim & Damian: Actions.

*Other issues*
    None.

*Adjournment*

    10:02 AM PDT

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

*Action items to be carried over*

      None

*New action items*

    Jim: Amend the nextafterd proposal with language about the
breaking-implementations issue, share with CFP, and then submit if there's
no objection.

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

    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.

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

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

    Jim: Draft change to references to rounding precision mode. Circulate
to new ad hoc rounding precision mode subcommittee: David, Damian, Jerome,
Jim.

*Discussion issues to be carried over*
     hypot(x, y) - C2Y F.10.5.4

    Occurs vs shall occur for math errors

    "result" vs. "result value" vs. "return value"

    SNAN macro location change

    hypot(x, NAN) in Annex F

    F.10.1#14 editorial issue

    canonicalize() in AnnexF - F.10.9.7

    Consistency in paragraphs using NaNs

    Preferred Style in Annex F - fromfp/fromfpx or nearbyint/rint

    Sign of NaN not interpreted

    Editorial issues with F.3#6

*Signoff*
    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/20251031/d63eab16/attachment.htm>