[Motions 2026 03 lwg 1] P4145R0 C++ Standard Library Ready Issues by burblebee · Pull Request #8926 · cplusplus/draft

burblebee · 2026-04-07T10:20:03Z

Also fixes cplusplus/nbballot#708
Also fixes cplusplus/nbballot#814
Also fixes cplusplus/nbballot#879
Also fixes cplusplus/nbballot#855
Also fixes cplusplus/nbballot#830

Notes:

LWG4474: See FIXMEs for concerns about the wording.

…al and variant

…ained by const Pattern

[atomics.ref.ops]p3 Added "," to match wording added in p1.

jensmaurer · 2026-04-07T11:31:32Z

I have added "fixes cplusplus/nbballot" links in the initial comment of this pull request. It would be good to amend the corresponding commits with our usual "Fixes NB ... (C++26 CD)." text (examples all over "git log").

Fixes NB GB 05-129 (C++26 CD).

…ecified

…only concept duplication Fixes NB US 151-242 (C++26 CD).

…r parallel algorithms

…the difference type is also a sentinel type

…ng (data_handle_type, mapping_type, accessor_type) and the corresponding constructor

…() member functions of std::filesystem::path are misnamed Fixes NB US 189-304 (C++26 CD).

Fixes NB US 176-280 (C++26 CD).

…ctions

[tab:time.format.spec],[tab:time.parse.spec] Fixed wording and formatting in examples

…:format

…td::basic_string on some platforms

…[] specification

Fixes NB US 157-255 (C++26 CD).

…nstant_array and related functions

Fixes NB US 152-243, PL-008 (C++26 CD).

… missing

…frob_norm

…elds

…n mismatch

frederick-vs-ja · 2026-04-09T05:21:10Z

source/support.tex

+if the rounding style is to the nearest representable value;
+%FIXME: The wording "equally near" seems problematic.
+if there are two equally near such values,
+%FIXME: Can multiple values have an "even least significant digit"? Or none at all?


I believe there's at least one such value. Although when the radix of the floating-point type is odd (which is theoretically permitted but doesn't seem present in any implementation), there may be two.

The wording is fine; I don't see any problem with the term "equally near". I could have phrased it as "equidistant" but that seemed too fancy.

An obvious example of where the "to even" part kicks in is when you divide the smallest possible float value by two. Zero and the subnormal minimum are equally near, and you disambiguate by choosing the number with the even least significant digit, which is zero. You get the same ambiguity halfway between all adjacent numbers with equal exponent at least.

The really interesting case is when even the "to even" disambiguation isn't sufficient 😄 I'm not sure about odd radix, but for even radix, this only occurs for floating-point types with a single digit mantissa digit. This case is deliberately left as a wording hole though in order to not mandate more than C's FLT_ROUNDS mandates. Implementers know what to do anyway.

If there can be more than 1 "equally near" values with "even least significant digit"s, then we can't chose "the one with an even least significant digit".

Okay yeah that's true. I think you have to explicitly spell out the ambiguity instead of acting as if there was none for this not to be a wording defect. That belongs outside this PR and either in a follow-up editorial fix or LWG issue though, unless the fix is really obvious, which it isn't.

If you wanted to be explicit about the remaining ambiguity, you would either need to go full roundTiesToEven as per ISO/IEC 60559 and disambiguate according to that (stricter than C), or leave it implementation-defined (imposes documentation requirement), or unspecified (may be non-deterministic), and which one to choose isn't remotely obvious or editorial.

jwakely · 2026-04-09T17:47:42Z

The fourth commit in the series (fd289eb) says "CWG" instead of "LWG" in the commit log.

jwakely · 2026-04-09T18:02:01Z

source/time.tex

 of offset from \tcode{\%EC} (year only).
+\begin{tailexample}
+\begin{codeblock}
+cout << format("{:%C %y}", -1976y);     // prints \tcode{-20 76}


The issue didn't ask for a codeblock, so this is an editorial change.

jwakely · 2026-04-09T18:02:31Z

source/time.tex

 the locale's alternative representation.
+\begin{tailexample}
+\begin{codeblock}
+year y;


This is also an editorial change, the issue didn't ask for a codeblock.

jwakely · 2026-04-09T20:00:13Z

source/containers.tex

 \rSec4[mdspan.sub.canonical]{\tcode{submdspan} slice canonicalization}

-\indexlibraryglobal{submdspan_canonicalize_slices}%
+%FIXME: "Slices" is used instead of "SliceSpecifiers" in the synopsis declaration.


This should be filed as an editorial issue on github, rather than as a FIXME comment in the source.

jwakely · 2026-04-09T20:09:17Z

source/meta.tex

 Let $P$ be
 \begin{itemize}
 \item
  the template parameter object\iref{temp.param}


I was confused why the surrounding context didn't match what was in the issue, but it was changed by a091468

jwakely

I added a few minor comments but everything looks correct. Nice work, Dawn!

burblebee added 6 commits April 6, 2026 16:32

LWG2414 Member function reentrancy should be implementation-defined

d5b6323

LWG2746 Inconsistency between requirements for emplace between option…

66f3f6e

…al and variant

LWG3504 condition_variable::wait_for is overspecified

f459b24

CWG3599 The const overload of lazy_split_view::begin should be constr…

fd289eb

…ained by const Pattern

LWG4290 Missing Mandates clauses on is_sufficiently_aligned

3262916

LWG4453 atomic_ref<cv T>::required_alignment should be the same as for T

2260b2a

[atomics.ref.ops]p3 Added "," to match wording added in p1.

burblebee force-pushed the motions-2026-03-lwg-1 branch from c3db190 to 1ec13a4 Compare April 7, 2026 13:21

burblebee marked this pull request as ready for review April 7, 2026 13:21

burblebee marked this pull request as draft April 7, 2026 13:47

burblebee added 20 commits April 7, 2026 06:53

LWG4454 assert should forbid co_await and co_yield

e8dfb74

Fixes NB GB 05-129 (C++26 CD).

LWG4469 Names of parameters of addressable function shall remain unsp…

c4a7293

…ecified

LWG4472 std::atomic_ref<const T> can be constructed from temporaries

b5910b5

LWG4486 integral-constant-like and constexpr-wrapper-like exposition-…

fe6d73c

…only concept duplication Fixes NB US 151-242 (C++26 CD).

LWG4492 std::generate and std::ranges::generate wording is unclear fo…

6662593

…r parallel algorithms

LWG4496 Precedes vs Reachable in [meta.reflection]

32bf976

LWG4499 flat_set::insert_range specification may be problematic

a8c8631

LWG4510 Ambiguity of std::ranges::advance and std::ranges::next when …

ecf1b99

…the difference type is also a sentinel type

LWG4511 Inconsistency between the deduction guide of std::mdspan taki…

cc98777

…ng (data_handle_type, mapping_type, accessor_type) and the corresponding constructor

LWG4512 The system_encoded_string() and generic_system_encoded_string…

2123003

…() member functions of std::filesystem::path are misnamed Fixes NB US 189-304 (C++26 CD).

LWG4535 Disallow user specialization of <simd> templates

cd9953e

Fixes NB US 176-280 (C++26 CD).

LWG4536 Type traits have inconsistent interactions with immediate fun…

7fed466

…ctions

LWG3831 Two-digit formatting of negative year is ambiguous

a102476

[tab:time.format.spec],[tab:time.parse.spec] Fixed wording and formatting in examples

LWG4090 Underspecified use of locale facets for locale-dependent std:…

3cdb19b

…:format

LWG4130 Preconditions for std::launder might be overly strict

865861d

LWG4259 P1148R0 changed the return values of searching functions of s…

bdffae9

…td::basic_string on some platforms

LWG4324 unique_ptr<void>::operator* is not SFINAE-friendly

9e936ca

LWG4378 Inconsistency between std::basic_string's data() and operator…

074f180

…[] specification

LWG4457 freestanding for stable_sort, stable_partition and inplace_merge

eba8f61

Fixes NB US 157-255 (C++26 CD).

LWG4460 Missing Throws: for last variant constructor

c4bab84

burblebee added 18 commits April 7, 2026 06:55

LWG4467 hive::splice can throw bad_alloc

7138a1d

LWG4468 [const.wrap.class] "operator decltype(auto)" is ill-formed

a675875

LWG4474 "round_to_nearest" rounding mode is unclear

bc2c669

LWG4477 Placement operator delete should be constexpr

6c9af3d

LWG4480 <stdatomic.h> should provide ATOMIC_CHAR8_T_LOCK_FREE

edc506e

LWG4481 Disallow chrono::duration<const T, P>

7bcf512

LWG4483 Multidimensional arrays are not supported by meta::reflect_co…

6da1031

…nstant_array and related functions

LWG4491 Rename submdspan_extents and submdspan_canonicalize_slices

ddefdcd

Fixes NB US 152-243, PL-008 (C++26 CD).

LWG4493 Specification for some functions of bit reference types seems…

ab81391

… missing

LWG4500 constant_wrapper wording problems

907d064

LWG4514 Missing absolute value of init in vector_two_norm and matrix_…

7b4c8c9

…frob_norm

LWG4517 data_member_spec should throw for cv-qualified unnamed bit-fi…

7dbbd7e

…elds

LWG4522 Clarify that std::format transcodes for std::wformat_strings

4279714

LWG4523 constant_wrapper should assign to value

6e26db0

LWG4525 task's final_suspend should move the result

30a71cc

LWG4527 await_transform needs to use as_awaitable

8a67cca

LWG4528 task needs get_completion_signatures()

0f900e8

LWG4529 task::promise_type::await_transform declaration and definitio…

076fc37

…n mismatch

burblebee force-pushed the motions-2026-03-lwg-1 branch from 1ec13a4 to 076fc37 Compare April 7, 2026 13:56

burblebee marked this pull request as ready for review April 7, 2026 13:58

frederick-vs-ja reviewed Apr 9, 2026

View reviewed changes

jwakely reviewed Apr 9, 2026

View reviewed changes

jwakely approved these changes Apr 9, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[Motions 2026 03 lwg 1] P4145R0 C++ Standard Library Ready Issues#8926

[Motions 2026 03 lwg 1] P4145R0 C++ Standard Library Ready Issues#8926
burblebee wants to merge 44 commits intomainfrom
motions-2026-03-lwg-1

burblebee commented Apr 7, 2026 •

edited by jensmaurer

Loading

Uh oh!

jensmaurer commented Apr 7, 2026

Uh oh!

frederick-vs-ja Apr 9, 2026

Uh oh!

eisenwave Apr 9, 2026 •

edited

Loading

Uh oh!

burblebee Apr 9, 2026

Uh oh!

eisenwave Apr 9, 2026 •

edited

Loading

Uh oh!

jwakely commented Apr 9, 2026

Uh oh!

jwakely Apr 9, 2026

Uh oh!

jwakely Apr 9, 2026

Uh oh!

jwakely Apr 9, 2026

Uh oh!

jwakely Apr 9, 2026

Uh oh!

jwakely left a comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

Conversation

burblebee commented Apr 7, 2026 • edited by jensmaurer Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jensmaurer commented Apr 7, 2026

Uh oh!

frederick-vs-ja Apr 9, 2026

Choose a reason for hiding this comment

Uh oh!

eisenwave Apr 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

burblebee Apr 9, 2026

Choose a reason for hiding this comment

Uh oh!

eisenwave Apr 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jwakely commented Apr 9, 2026

Uh oh!

jwakely Apr 9, 2026

Choose a reason for hiding this comment

Uh oh!

jwakely Apr 9, 2026

Choose a reason for hiding this comment

Uh oh!

jwakely Apr 9, 2026

Choose a reason for hiding this comment

Uh oh!

jwakely Apr 9, 2026

Choose a reason for hiding this comment

Uh oh!

jwakely left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

burblebee commented Apr 7, 2026 •

edited by jensmaurer

Loading

eisenwave Apr 9, 2026 •

edited

Loading

eisenwave Apr 9, 2026 •

edited

Loading