[Motions 2026 03 lwg 2] P4146R0 C++ Standard Library Immediate Issues by burblebee · Pull Request #8935 · cplusplus/draft

burblebee · 2026-04-09T14:05:23Z

Fixes #8836.

Fixes NB RU 268 (C++26 CD).

…e first range is input and one of the ranges is empty

…lity Fixes NB US 135-216, US 136-217 (C++26 CD).

…g std::inplace_vector

…s return void

…types

Fixes NB US 242-372 (C++26 CD).

Fixes NB US 257-382 (C++26 CD). [task.state]p5 Didn't insert a new paragraph (text belongs with the \effects clause).

…arantee

…constraints

… min/max considerations Fixes NB US 142-236 (C++26 CD).

Fixes NB US 249-379 (C++26 CD).

… requests to spawned work [exec.spawn.future] "receiver" in struct "future-operation" renamed to "receiverx" due to concept of same name.

…in_out_result

[set.intersection] Turned "skipped" into an indexed definition to match that of LWG4544.

Fixes NB GB 07, GB 11 (C++26 CD).

… memory contents

…to concat_view Fixes NB FR 025-246 (C++26 CD).

Fixes NB US 139-232 (C++26 CD).

Fixes NB US 81-149 (C++26 CD).

Fixes NB US 75-138 (C++26 CD).

jwakely · 2026-04-14T12:05:35Z

source/utilities.tex

 is \tcode{true}.
 \end{itemize}
+
+%FIXME: Is this supposed to be part of the Remarks clause?


Yes, but it's OK as a separate paragraph that follows the existing Remarks: paragraph (as discussed on mattermost)

jwakely · 2026-04-14T12:18:34Z

source/containers.tex

  using @\exposid{range-to-alloc-type}@ =
-    pair<const typename ranges::range_value_t<Range>::first_type,
-         typename ranges::range_value_t<Range>::second_type>;           // \expos
+    pair<const @\exposid{range-key-type}@<Range>,


N.B. The P/R in the issue shows add_const_t here, but that was changed to const by #7851 so the conflict here has been resolved correctly.

jwakely

Some changes, some comments not requiring actions, and some questions.

jwakely · 2026-04-14T12:35:14Z

source/exec.tex

-Let \tcode{\placeholder{st}} be \tcode{get_stop_token(get_env(\exposid{rcvr}))}.
-Initializes \tcode{\placeholder{prom}.\exposid{token}} and
-\tcode{\placeholder{prom}.\exposid{source}} such that
+After that invokes \tcode{\exposid{handle}.resume()}.


I like the change to use "Finally, returns ..." below, would it make sense to be consistent here?

Suggested change

After that invokes \tcode{\exposid{handle}.resume()}.

Finally, invokes \tcode{\exposid{handle}.resume()}.

jwakely · 2026-04-14T12:37:47Z

source/exec.tex

+After that invokes \tcode{\exposid{handle}.resume()}.
+\end{itemdescr}
+
+\indexlibrarymember{\exposid{get-stop-token}}{task::\exposid{state}}%


Should this exposition-only member function be indexed? Huh, we do it for lots of others ... OK then.

jwakely · 2026-04-14T12:38:31Z

source/exec.tex

+\effects
+If
+\tcode{\libconcept{same_as}<decltype(declval<stop_source_type>().get_token()), decltype(get_\linebreak{}stop_token(get_env(\exposid{rcvr})))>}
+is \tcode{true}


Suggested change

is \tcode{true}

is \tcode{true},

jwakely · 2026-04-14T12:43:17Z

source/exec.tex

 Otherwise, it is expression-equivalent to
 \tcode{\exposid{MANDATE-NOTHROW}(rcvr.set_value(vs...))}.

+\mandates


Do we want a new paragraph for this? The issue doesn't request it, and some other places in [exec] have Mandates attached to other paragraphs (or list items), so I'm not sure.

Suggested change

\mandates

\pnum

\mandates

If we do it here, we should do it below as well.

(Maybe our library element macros like \mandates and \expects should have an implicit \pnum, and then have a raw form for cases like this and [structure.specifications] and [exec.get.fwd.progress] p2 where we don't want a new paragraph and a pnum.)

jwakely · 2026-04-14T13:07:12Z

source/exec.tex

+    using @\exposid{stop-callback-t}@ =                                     // \expos
+      stop_callback_for_t<@\exposid{stop-token-t}@, @\exposid{callback}@>;
+
+    struct @\exposid{receiverx}@ {                                          // \expos


I don't love just appending 'x' to disambiguate this from the receiver concept. Is it a problem because we have scripts that complain that \libconcept is not used on it? Could we fix the scripts to ignore \exposid{foo} so we can reuse the name? Or maybe make this receiver-t instead? That's not ideal because we also have receiver_t but at least they're not identical.

jwakely · 2026-04-14T13:09:48Z

source/exec.tex

+    @\exposid{future-operation}@(StatePtr state, Rcvr rcvr) noexcept        // \expos
+      : @\exposid{rcvr}@(std::move(rcvr)), @\exposid{state}@(std::move(state)), @\exposid{inner}@(this)


Do we care about the shadowed named here?

Suggested change

@\exposid{future-operation}@(StatePtr state, Rcvr rcvr) noexcept // \expos

: @\exposid{rcvr}@(std::move(rcvr)), @\exposid{state}@(std::move(state)), @\exposid{inner}@(this)

@\exposid{future-operation}@(StatePtr s, Rcvr r) noexcept // \expos

: @\exposid{rcvr}@(std::move(r)), @\exposid{state}@(std::move(s)), @\exposid{inner}@(this)

I don't think we need to change anything, it's fine as is.

jwakely · 2026-04-14T13:12:23Z

source/exec.tex

+        }
+      }
+
+      @\exposid{state}@.release()->@\exposid{consume}@(@\exposid{innex}@);


Suggested change

@\exposid{state}@.release()->@\exposid{consume}@(@\exposid{innex}@);

@\exposid{state}@.release()->@\exposid{consume}@(@\exposid{inner}@);

jwakely · 2026-04-14T13:18:06Z

source/algorithms.tex

-are considered \term{skipped}.
-If $N < M$, a non-copied element is also considered skipped
-if it compares less than the $(N + 1)^\text{th}$ element
+\indextext{element!skipped}%


This definition is local to this function, do we need to index it?

jwakely · 2026-04-14T13:19:22Z

source/algorithms.tex

 are included in the sorted difference, in order.
+Of those equivalent elements,
+the first $\min(m, n)$ elements in both ranges
+\indextext{element!skipped}%


Same question here, do we need to index this?

jwakely · 2026-04-14T13:45:04Z

source/meta.tex

 returns an object \exposid{V} such that
 \tcode{\exposid{V}.data()[\exposid{V}.size()]} equals \tcode{'\textbackslash 0'}.
+Each element of the range
+\tcode{\exposid{V}.data() +} \range{0}{\exposid{V}.size()}


This should be a closed range, not half-open, and the + should be math font not code font:

Suggested change

\tcode{\exposid{V}.data() +} \range{0}{\exposid{V}.size()}

$\tcode{\exposid{V}.data()} + \crange{0}{\exposid{V}.size()}$

We have the \countedrange macro for X+[0,n) but it always uses a half-open range. We could add \countedcrange for X+[0,n] but I think this is the only place we need it.

burblebee marked this pull request as ready for review April 9, 2026 14:26

eisenwave added this to the post-2026-03 milestone Apr 12, 2026

tkoeppe force-pushed the motions-2026-03-lwg-2 branch from 29c2e35 to ed9db42 Compare April 12, 2026 20:19

burblebee added 8 commits April 12, 2026 21:19

LWG3662 basic_string::append/assign(NTBS, pos, n) suboptimal

9d7581a

Fixes NB RU 268 (C++26 CD).

LWG3777 Common cartesian_product_view produces an invalid range if th…

581d0cb

…e first range is input and one of the ranges is empty

LWG3797 elements_view insufficiently constrained

a50b69c

LWG3891 LWG 3870 breaks std::expected<cv T, E>

4fa9e57

LWG4026 Assignment operators of std::expected should propagate trivia…

0b97fac

…lity Fixes NB US 135-216, US 136-217 (C++26 CD).

LWG4122 Ill-formed operator<=> can cause hard error when instantiatin…

9555e19

…g std::inplace_vector

LWG4133 awaitable-receiver's members are potentially throwing

d323c5a

LWG4143 execution::set_value/set_error/set_stopped/start should alway…

901b485

…s return void

tkoeppe force-pushed the motions-2026-03-lwg-2 branch from ed9db42 to 45d9c2d Compare April 12, 2026 20:21

burblebee added 18 commits April 13, 2026 23:42

LWG4151 Precondition of inplace_vector::swap

8f66474

LWG4223 Deduction guides for maps are mishandling tuples and references

ca72aac

LWG4314 Missing move in mdspan layout mapping::operator()

0514f2f

LWG4325 std::indirect's operator== still does not support incomplete …

9ad14eb

…types

LWG4339 task's coroutine frame may be released late

1be5f6e

Fixes NB US 242-372 (C++26 CD).

LWG4347 task's stop source is always created

c4ccbb2

Fixes NB US 257-382 (C++26 CD). [task.state]p5 Didn't insert a new paragraph (text belongs with the \effects clause).

LWG4354 Reconsider weakly_parallel as the default forward_progress_gu…

e85c4b0

…arantee

LWG4361 awaitable-receiver::set_value should use Mandates instead of …

c8b37c1

…constraints

LWG4379 hive::reserve() needs Throws: element adjusted to match block…

6fe0756

… min/max considerations Fixes NB US 142-236 (C++26 CD).

LWG4476 run_loop should not have a set_error completion

1743833

LWG4478 meta::has_identifier is not specified for annotations

073a3ed

LWG4485 Move specification for task::stop_token_type

c2be2be

Fixes NB US 249-379 (C++26 CD).

LWG4497 std::nullopt_t should be comparable

140c4a0

LWG4504 Wording problem in {simple_}counting_scope

0922696

LWG4506 source_location is explicitly unspecified if is constexpr or not

415d2ac

LWG4532 Imprecise std::polymorphic wording seems to imply slicing

bc841ac

LWG4533 not_fn<f> is unimplementable

8336444

LWG4537 Improve define_static_array

b36dfac

burblebee added 11 commits April 13, 2026 23:42

LWG4540 future-senders returned from spawn_future do not forward stop…

4ae4a80

… requests to spawned work [exec.spawn.future] "receiver" in struct "future-operation" renamed to "receiverx" due to concept of same name.

LWG4544 Parallel overload of ranges::set_difference should return in_…

5a6ac9c

…in_out_result

LWG4548 Parallel ranges::set_intersection should not do unnecessary work

171ed2c

[set.intersection] Turned "skipped" into an indexed definition to match that of LWG4544.

LWG4549 vprint_nonunicode_buffered ignores its stream parameter

c4ad4af

LWG4550 Need new feature test macros for <stdckdint.h> and <stdbit.h>

a1901ce

Fixes NB GB 07, GB 11 (C++26 CD).

LWG4552 compare_exchange_weak writes a value on spurious failure, not…

b08f0d6

… memory contents

LWG4553 Wording for FR-025-246 25.7.18.2 Add a reserve_hint function …

c4e0c92

…to concat_view Fixes NB FR 025-246 (C++26 CD).

LWG4554 Remove undefined behaviour from hive for invalid limits

79358bc

Fixes NB US 139-232 (C++26 CD).

LWG4555 Remove is_consteval_only

7738975

Fixes NB US 81-149 (C++26 CD).

LWG4556 Unclear properties of reflection strings

1e44b03

LWG4557 Remove constexpr from owner_less and owner_before

b87b2ae

Fixes NB US 75-138 (C++26 CD).

tkoeppe force-pushed the motions-2026-03-lwg-2 branch from 45d9c2d to b87b2ae Compare April 13, 2026 22:43

jwakely reviewed Apr 14, 2026

View reviewed changes

jwakely requested changes Apr 14, 2026

View reviewed changes

	After that invokes \tcode{\exposid{handle}.resume()}.
	Finally, invokes \tcode{\exposid{handle}.resume()}.

		@\exposid{future-operation}@(StatePtr state, Rcvr rcvr) noexcept // \expos
		: @\exposid{rcvr}@(std::move(rcvr)), @\exposid{state}@(std::move(state)), @\exposid{inner}@(this)

	@\exposid{state}@.release()->@\exposid{consume}@(@\exposid{innex}@);
	@\exposid{state}@.release()->@\exposid{consume}@(@\exposid{inner}@);

	\tcode{\exposid{V}.data() +} \range{0}{\exposid{V}.size()}
	$\tcode{\exposid{V}.data()} + \crange{0}{\exposid{V}.size()}$

Conversation

burblebee commented Apr 9, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jwakely left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jwakely Apr 14, 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!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

jwakely Apr 14, 2026 •

edited

Loading