[Cfp-interest 3307] JTC1/SC22/WG14/CFP 2024/09/18 Meeting minutes v2
Jerome Coonen
jcoonen at gmail.com
Wed Oct 9 11:59:44 PDT 2024
[These minutes appear in agenda order, with carry-over and new agenda items
appearing at the end.]
2024/09/18: 8 AM PDT
*Attendees*: Rajan, Jim, Jerome, Fred, Damian, David, Joshua
*New agenda items* (
https://wiki.edg.com/pub/CFP/WebHome/CFP%20meeting%20agenda-20240918-update.pdf
):
SIGFPE
[Cfp-interest 3264] SIGFPE paper Fred J. Tydeman
*Previous meeting notes:*
See([Cfp-interest 3253] JTC1/SC22/WG14/CFP 2024/08/14 Meeting minutes
-- v. 2
<http://%5BCfp-interest%203253%5D%20JTC1/SC22/WG14/CFP%202024/08/14%20Meeting%20minutes%20--%20v.%202>
).
*Next Meeting(s):*
October 16, 2024, 3PM UTC
ISO Zoom teleconference
Please notify the group if this time slot does not work.
*IEEE 754 liaison:*
Damian & David: Large, newly-formed committee working on how to
expedite discussion of minor issues without the inevitable distractions.
*C++ liaison:*
None
*WG14 meeting:*
30 Sept - 4 Oct in Minneapolis, MN
*C23 integration:*
Mailing deadline is one month before the meeting, given above, so it
has passed.
*C documents:*
The C23 DIS draft is n3219 - 22 Feb 2024
The working C2y draft is n3301 - 28 July 2024 - For CFP review only. Do
not distribute.
*Carry-over action items from last meeting:*
Damian: Ask 754 about midpoint and interpolation
Damian: Will send email to 754, to carry over
Jim: Include complex add and sub into the main body.
Jim & everyone: Look at ways to strengthen the complex * and /, given
the current code block.
[Cfp-interest 3155] Specification of complex operators --Done
Jim: Write up notes about frexp() and behavior for model numbers
[Cfp-interest 3154] frexp and double-double --Done
*Action items from last meeting:*
All: Investigate issue [Cfp-interest 3196] Fwd: [SC22WG14.26243]
constexpr initialization seemingly can contai UB?I from email forwarded by
Fred, involving a constant expression of the form 1.0/0.0, for example.
--Done
All: Investigate issue [Cfp-interest 3195] Fwd: [SC22WG14.26260] Can
nan set errno? from an email forwarded by Fred, involving the possible
setting of errno by the nan() functions. --Done
Jim: Add 2 issues raised by Rajan regarding WG14. --Done
Jim: Add actions for group in short term. --Done
Jim: Draft proposal to remove instances of imaginary lingering in
draft. --Done
[Cfp-interest 3250] Lingering references to imaginary type
Jerome: add document # to pole proposal. --Done
N3324 2024/09/04 Thomas, C2Y proposal - wording for pole error
--Done
Fred: investigate range error issue for llogb(), which differs from
ilogb(). --Done
[Cfp-interest 3202] llogb() Fred J. Tydeman
Jerome: Identify fadd/fsub and fma "error may occur" issues and report.
--carry over
Damian: Send email about sign conventions in Annex F & G proposal.
--Done
[Cfp-interest 3200] Various Updates to Annex F+G
[Cfp-interest 3203] Re: Various Updates to Annex F+G
Damian & Jerome: Identify technical changes in the F & G proposal.
Investigate ways to split the proposal into natural subsets. ---Done
Fred: Revise the proposal for SIGFPE and I/O, in light of further
discussion --Done
[Cfp-interest 3264] SIGFPE paper
*TS-4 and TS-5 revisions*
Waiting to hear from ISO editors, or word that the TSes (as we
submitted them) have been sent out to ballot.
*C2y issues*
None.
*Discussion of issues*
Moving complex ops spec from Annex G to main body
[Cfp-interest 3262] Re: Specification of complex operators Jim
Thomas
[Cfp-interest 3263] Re: Specification of complex operators Damian
McGuckin
[Cfp-interest 3267] Re: Specification of complex operators Jim
Thomas
[Cfp-interest 3272] Re: Specification of complex operators Damian
McGuckin
[Cfp-interest 3273] Re: Specification of complex operators Damian
McGuckin
[Cfp-interest 3278] Re: Specification of complex operators Paul
Zimmermann
[Cfp-interest 3275] Re: Specification of complex operators Jim
Thomas
Jim: Discussed paper close to ready to be a proposal. Propose
to add more substantive discussion than current table of operations.
Introduce "x + I y" usage. Explain complex mul, avoiding "undue"
exceptions, leading to other operations. This replaces the tables.
Fred: Won't cover cases involving IEEE signed 0, signed
infinity, and NaN. Suggest at least a footnote about special values.
Rajan: ...not necessarily just IEEE values.
Joshua: Not sure "compute" is best word. Like the idea of
specifying what is wanted, even without all the details.
David: "Evaluates"?
Jim: Also a tricky word. Decide to take another turn and
discuss again.
Jim: Clarify the usage of "I"
frexp and double-double
[Cfp-interest 3100] Issue: frexp and double-double underflow Hubert
Tong
[Cfp-interest 3101] Re: Issue: frexp and double-double underflow
Fred J. Tydeman
[Cfp-interest 3102] Re: Issue: frexp and double-double underflow
Fred J. Tydeman
[Cfp-interest 3103] Re: Issue: frexp and double-double underflow
Hubert Tong
[Cfp-interest 3104] Re: [SC22WG14.25365] Issue: frexp and
double-double underflow Hubert Tong
[Cfp-interest 3154] frexp and double-double Jim Thomas
Jim: Review issue of double double. Two suggestions. One lets
result be unspecified if value not a model number. Other path is a dead end.
David: Say as little as possible.
All: agree to first option
constexper initialization with UB
[Cfp-interest 3196] Fwd: [SC22WG14.26243] constexpr initialization
seemingly can contain UB? Fred J. Tydeman
[Cfp-interest 3225] Re: [SC22WG14.26243] constexpr initialization
seemingly can contain UB? Jim Thomas
[Cfp-interest 3226] Re: [SC22WG14.26243] constexpr initialization
seemingly can contain UB? Fred J. Tydeman
Jim: Discuss proposed language for UB
Rajan: WG14 seems to want more language, but can try this
first
nan function and errno
[Cfp-interest 3195] Fwd: [SC22WG14.26260] Can nan set errno? Fred
J. Tydeman
[Cfp-interest 3198] Re: Can nan() set errno? Fred J. Tydeman
[Cfp-interest 3210] Re: Can nan() set errno? Jim Thomas
[Cfp-interest 3227] Re: Can nan() set errno? Fred J. Tydeman
[Cfp-interest 3229] Re: Can nan() set errno? Jim Thomas
[Cfp-interest 3230] Re: Can nan() set errno? Joshua Cranmer
[Cfp-interest 3237] Re: Can nan() set errno? Fred J. Tydeman
[Cfp-interest 3238] Re: Can nan() set errno? Jim Thomas
[Cfp-interest 3239] Re: Can nan() set errno? Fred J. Tydeman
[Cfp-interest 3256] Re: Can nan() set errno? Jim Thomas
[Cfp-interest 3259] Re: Can nan() set errno? Fred J. Tydeman
Jim: Tied to strtod()
Jerome: Should be simple, quiet NaN value, not an exotic
implementation-defined value with extra functionality.
Joshua: 754 converts an integer to a NaN
Fred: Could change strtod() to specify domain errors for
"unusual" string
Jim: Should there be a note making strtod() behavior more
explicit? Will hold off for now.
mid-point and linear interpolation functions
[Cfp-interest 3145] midpoint function David Hough CFP
[Cfp-interest 3147] Re: midpoint function Vincent Lefevre
[Cfp-interest 3149] Re: midpoint function Mike Cowlishaw
[Cfp-interest 3151] Re: midpoint function Vincent Lefevre
[Cfp-interest 3150] Re: midpoint function David Hough CFP
[Cfp-interest 3146] thoughts on linear interpolations David Hough
CFP
[Cfp-interest 3148] Re: thoughts on linear interpolations Damian
McGuckin
Damian: initial disc, leading to Vincent's reply, then 3146
from David ("eloquent" even) arguing it might be overkill.
David: Hard to standardize a function when not an obvious best
way to accomplish it.
All: Sidebar on gamma and Bessel functions, arguably outside
scope of C
David: LAPACK has a def. but that doesn't mean the language
standards should.
Jim: Case where functions are "slow" in software yet unlikely
to be implemented in hardware.
Joshua: Added to C++ language standard, with many good
properties (though not perfect, with respect to IEEE exceptions, and
others). Will check how much it's used now.
David: Suggest matching C++ syntax, not semantics
Jim: After discussion, seems there is no need for a fully
specified version of the function.
lingering imaginary types
[Cfp-interest 3250] Lingering references to imaginary type Jim
Thomas
Jim: Proposal is ready to go.
All: As an additional item, the language "....return the
imaginary part value (as a real)" in the cimag functions feels redundant
(Jim) but there was support to keep the parenthetical for absolute clarity.
The issue needs more discussion.
use of “can not”
[Cfp-interest 3162] use of "can not" in N3301 5.2.2.4p13 Vincent
Lefevre
Jim: Issue is spilling a register without changing its value.
Suggest change to "cannot" to remove subtle ambiguity of the meaning of
"can not", in context
range errors for ilogb and llogb
[Cfp-interest 3202] llogb() Fred J. Tydeman
[Cfp-interest 3228] Re: llogb() Jim Thomas
[Cfp-interest 3266] Re: llogb() Jim Thomas
Jim: Current usage has over/underflow of just floating point
values. This conflicts with the language of ilogb() and llogb(). Suggest
removing all references to range error when the result is an integer type
Rajan: WG14 won't like "obsolescent features"
*Other issues:*
None.
*Adjournment*
10:04 AM PDT
-----------------------------------------------
*Action items to be carried over:*
Jerome: Identify fadd/fsub and fma "error may occur" issues and report.
*New action items:*
Jim: Update Complex suggestions (3262) for Annex G migration.
Jim: Update frexp for double double (3154) per discussion, to leave
undefined result for non-model numbers, and submit the proposal to WG14.
Fred: Send note to WG14 about constexper exceptional case issue.
Fred: Send nan() email to WG14 (3259).
David : Draft reply to WG14 re. midpoint and interpolation.
Jim: Submit to WG14 a proposal about lingering references to imaginary
(3250).
Fred: Draft a proposal for the "cannot" vs. "can not" change (3162).
All: Please send comments to Damian by Monday 30 Sept on the the first
of 4 sets of proposals (3247), including comments on Jim's responses (3279).
Jim: Write up a proposal for ilogb() and llogb() (3266).
*Discussion issues to be carried over:*
fadd/fsub and fma "error may occur" issues
treatment of error conditions
[Cfp-interest 3193] improving the language of 7.12.2 "Treatment
of error conditions" Jerome Coonen
changes to 6.X and 7.X
[Cfp-interest 3247] C26 Complex Related Changes Damian McGuckin
[Cfp-interest 3248] Re: C26 Complex Related Changes Damian
McGuckin
[Cfp-interest 3249] Re: C26 Complex Related Changes Jim Thomas
[Cfp-interest 3279] Re: C26 Complex Related Changes Jim Thomas
deeper TOC for annexes
[Cfp-interest 3164] TOC depth for C annexes
Respectfully submitted.
[v2 changes: Refine a comment by Fred; tweak the opening editorial comment.]
-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/20241009/ece0c1b8/attachment-0001.htm>
More information about the Cfp-interest
mailing list