[Cfp-interest 3333] JTC1/SC22/WG14/CFP 2024/11/20 Meeting minutes -- v. 2

Thu Nov 21 11:34:06 PST 2024

[Further comments on these minutes are welcome. This draft has a major
update of Joshua's C++ report, as well as changes re. SIGFP and complex.]

2024/11/20: 8 AM PDT

*Attendees*

    Jim, Jerome, Fred, Joshua, Damian, David

*New agenda items*

    https://wiki.edg.com/pub/CFP/WebHome/n3387.pdf

*Previous meeting notes*

    https://wiki.edg.com/pub/CFP/WebHome/n3386.pdf

*Next Meeting(s)*

    11 December 2024, 4PM UTC if needed

    8 January 2025, 4PM UTC

    ISO Zoom teleconference

     Please notify the group if these time slots do not work.

*C documents*

    The working C2y draft is n3301 - 28 July 2024 - For CFP review only. Do
not distribute.

*IEEE 754 liaison*

   Fred, Jerome: There is new discussion of decimal vs. binary comparisons
(among many other things).

*C++ liaison*

    Joshua: Meeting this week in Wroclaw, Poland. One issue is on the
FLT_EVAL_METHOD semantics in C++, which is currently largely based on C
semantics, but has unclear extension to C++ features (particularly
constexpr contexts). There is another set of issues about floating-point
reproducibility in C++, and there is a growing appetite to specify the
semantics more completely. Pragmas are not popular in C++. Templates are an
alternative, but not without issues.

[Also added in edit, Joshua offers to write up a more thorough summary of
the C++ floating-point discussions, if that is of interest. Finally, the
pronunciation of the name of the city many of us learned as Warsaw is
"VROWT-slaaf".]

*WG14*

    None

*C23 integration*

    C23 has been published. Has an October 2024 date, but referred to as
C23.

    https://www.iso.org/standard/82075.html

*Carry-over action items from last meeting *

    David: Draft reply to WG14 re midpoint and interpolation. Will carry
over.

*Action items from last meeting *

    Fred: Submit update to SIGFPE proposal N3378, with the suggested new
wording re. the raise function.

      Not submitted and action closed. Fred discovered sufficiently many
issues in preparing the proposal to warrant CFP review and a new action
item below.

    Jerome: Draft a proposal fadd, etc. (3314) following the discussion
above, and adhering to the practice of new text in green and removed text
in red strike-thru. Done.

https://wiki.edg.com/pub/CFP/WebHome/fadd%20fsub%20etc%20error%20conditions%20proposal.pdf

    Jerome: Draft proposal re. changes for "treatment of error con,
watching for interaction with Jim's change re. range error in 7.12.2#4. Add
a note to the editor that Jim's change below comes first, then this tweak
of the verb. Done

https://wiki.edg.com/pub/CFP/WebHome/treatment%20of%20error%20conditions%20proposal.pdf

    Damian: Split doc into 3 documents: complex.h intro, cproj, and
editorial suggestions. Will carry over.

    Jim: Submit proposal re. range errors for ilogb and llogb (3300). Done,
closed.
      https://www.open-std.org/jtc1/sc22/wg14/www/docs/n3383.pdf

    Jim: Submit proposal re moving complex operators spec from Annex G to
main body (3285), with an eye to fixing the rogue underscore-u references
that Fred spotted. Done, but revised into updated action item.
      https://www.open-std.org/jtc1/sc22/wg14/www/docs/n3384.pdf

*TS-4 and TS-5 revisions*

    Jim: New draft in cycle of edits arrived from ISO just minutes before
this meeting. More to come.

*Discussion of issues*

    SIGFPE

      https://www.open-std.org/jtc1/sc22/wg14/www/docs/n3378.htm

      Fred: Finds further references to SIGFPE in the standard, some of
which lead to undefined behavior. Fred lists 4 options for how to proceed
with SIGFPE and is in contact with WG14.

      https://wiki.edg.com/pub/CFP/WebHome/n3378c.htm

      Jim: One mitigating factor is that functions have other sources of
undefined behavior. Propose that Fred post to WG14 to gauge their sense of
limiting occurrences of SIGFPE.

    fadd/fsub and fma "error may occur" issues
      Jerome: Quick review of topic discussed at last meeting.

    Treatment of error conditions
     Jerome: Quick review.
     Fred: ISO "may" balk at "can" vs. "may" occur, but we can submit and
see how the editors reply.
    Jim: Seems to address a confusing issue that has prevailed for some
time.

    Changes to 6.X and 7.X
      Jim: Damian will send email just after the meeting for further
discussion next time.

    Moving complex operators spec from Annex G to main body
      https://www.open-std.org/jtc1/sc22/wg14/www/docs/n3384.pdf
      Jim: Elaborates on semantics discussion in 6.5.6, about computing A +
iB. Got comments from first distribution about awkward wording and use of
"should" in normative text. Discusses revised draft, with a Recommended
Practice statement. Well received.

    ERANGE question
      Jim: In 3303, Paul Zimmermann asked about huge values rounding down
to MAX. In 3305 Fred explained that ordinary rounding (vs. directed
rounding) could lead to MAX without an overflow range error. Jim reviews
the definition from 754. Double double complicates things, leading to much
discussion. Suggest taking no further action.

    Imaginary I
      https://www.open-std.org/jtc1/sc22/wg14/www/docs/n3390.htm
      Jim: Explains WG14 proposal to remove I macro. It is claimed that
with the i suffix that I is redundant. One matter is that without the
imaginary type, the macro must be defined as a complex.
      Damian: Could use CMPLX macro to define the I macro.
      Jim: Keep on the agenda for a closer look. One issue is when we might
see a draft with the imaginary i suffix and other approved changes related
to complex. The current language seems to allow the imaginary types as an
extension. Many details of recent changes must be checked if imaginary
types are allowed.

    i suffix warning
      Jim: Suggests adding a note warning about the use of the i suffix to
generate values with signed 0 and inf involved and advocating use of CMPLX
macros instead.
      Damian: It will be a forward reference (but there are plenty of them).

*Other issues:*

    None.

*Adjournment*

    9:37 AM PDT

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

*Action items to be carried over:*

    David: Draft reply to WG14 re midpoint and interpolation.

    Damian: Split doc into 3 documents: complex.h intro, cproj, and
editorial suggestions.

*New action items:*

    Fred: Submit SIGFPE suggestions to WG14 for comment.

    Jerome: Submit fadd, fsub, etc. domain errors proposal to WG14. (Will
require help from another member actually certified to submit to WG14.)

    Jerome: Submit exception "occurs" proposal to WG14 (again, with
support).

    Jim: Submit revised complex operators proposal to WG14.

    All: Continue study of the imaginary I macro and i suffix issues,
noting that the I macro may be removed from C.

    Jim: Write a proposal for use of CMPLX to generate desired values with
signed 0 and inf when using the i suffix in a context with no imaginary
values, just real and complex.

*Discussion issues to be carried over:*

    None.

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/20241121/8c61d470/attachment.htm>