Meta LWG issue: 2021-10 meeting

[StephanTLavavej]

(Previous meta-issue: #1965)

At the October 2021 virtual plenary meeting, the following LWG issues were resolved in the C++ Working Paper.

❔ Not yet analyzed

• Analyzed all issues! 😸

✖️ Not applicable

If an issue requires no action from implementers, we mark it as N/A. Categories:

• Pure wording clarifications with nothing to implement (these can be changes to non-normative text like examples and informative notes, or wording cleanups to normative text that don't impact observable behavior)
  • LWG-2191 Incorrect specification of match_results(match_results&&)
  • LWG-3123 duration constructor from representation shouldn't be effectively non-throwing
  • LWG-3152 common_type and common_reference have flaws in common
  • LWG-3361 safe_range<SomeRange&> case
  • LWG-3559 Semantic requirements of sized_range are circular
  • LWG-3573 Missing Throws element for basic_string_view(It begin, End end)
• Something that increases the restrictions placed on users, but implementers aren't expected to enforce those restrictions
• Fixes for obviously broken wording, where implementers would have done the right thing anyways

😸 Already implemented

Sometimes we cite LWG issues in product code comments as we're implementing their proposed resolutions. When the resolutions are officially accepted, we should remove the citations (as the default assumption is that we're implementing what the Standard says). If something is especially subtle, we can convert the citation to mention the relevant Standard section.

Sometimes we should add test coverage - e.g. when the Standard begins requiring something that we were already doing, but weren't explicitly testing for.

• Already implemented, comments need to be removed and messages need to cite the Standard
• Implemented without comments
  • LWG-2762 unique_ptr operator*() should be noexcept (implemented in implemented the proposed resolution for LWG-2762 #2376)
  • LWG-3293 move_iterator operator+() has incorrect constraints (comments removed by Remove LWG comments #2269)
  • LWG-3407 Some problems with the wording changes of P1739R4 (comments removed by Remove LWG comments #2269)
  • LWG-3422 Issues of seed_seq's constructors (implemented in LWG-3422: Issues of seed_seq's constructors #2260)
  • LWG-3480 directory_iterator and recursive_directory_iterator are not C++20 ranges (implemented in Implement LWG-3480, directory_iterator and recursive_directory_iterator are ranges #2340)
  • LWG-3498 Inconsistent noexcept-specifiers for basic_syncbuf (comments removed by Remove LWG comments #2269)
  • LWG-3535 join_view::iterator::iterator_category and ::iterator_concept lie (comments removed by Remove LWG comments #2269)
  • LWG-3554 chrono::parse needs const charT* and basic_string_view<charT> overloads (implemented by LWG-3554 Add const charT* overloads for chrono::parse #2261)
  • LWG-3560 ranges::equal and ranges::is_permutation should short-circuit for sized_ranges (implemented in the original PRs Several range algorithms #565 and Implement ranges::is_permutation #892)
  • LWG-3566 Constraint recursion for operator<=>(optional<T>, U) (comments removed by Remove LWG comments #2269)
  • LWG-3567 Formatting move-only iterators, take two (implemented in original PR Merging format #1821)
  • LWG-3568 basic_istream_view needs to initialize value_ (comments removed by Remove LWG comments #2269)
  • LWG-3572 copyable-box should be fully constexpr (implemented in Implement P2325R3 Views Should Not Be Required To Be Default Constructible #2012)
  • LWG-3581 The range constructor makes basic_string_view not trivially move constructible (comments removed by Remove LWG comments #2269)
  • LWG-3585 variant converting assignment with immovable alternative

🩹 Patches an unimplemented feature

We should record this LWG issue in the GitHub issue tracking the feature. That way, we'll remember to verify it, but it doesn't represent net new work.

🐞 Not yet implemented

• Filed a GitHub issue labeled LWG
  • LWG-2381 Inconsistency in parsing floating-point numbers (filed LWG-2381 Inconsistency in parsing floating point numbers #2388)
  • LWG-3121 tuple constructor constraints for UTypes&&... overloads (filed LWG-3121 tuple constructor constraints for UTypes&&... overloads #2390)
  • LWG-3146 Excessive unwrapping in std::ref/cref (filed LWG-3146 Excessive unwrapping in std::ref/cref #2391)
  • LWG-3392 ranges::distance() cannot be used on a move-only iterator with a sized sentinel (filed LWG-3392 ranges::distance() cannot be used on a move-only iterator with a sized sentinel #2392)
  • LWG-3470 convertible-to-non-slicing seems to reject valid case (filed LWG-3470 convertible-to-non-slicing seems to reject valid case #2393)
  • LWG-3557 The static_cast expression in convertible_to has the wrong operand (filed LWG-3557 The static_cast expression in convertible_to has the wrong operand #2394)
  • LWG-3561 Issue with internal counter in discard_block_engine (filed LWG-3561 Issue with internal counter in discard_block_engine #2250)
  • LWG-3563 keys_view example is broken (filed LWG-3563 keys_view example is broken #2389)
  • LWG-3570 basic_osyncstream::emit should be an unformatted output function (filed LWG-3570 basic_osyncstream::emit should be an unformatted output function #2395)
  • LWG-3571 flush_emit should set badbit if the emit call fails (filed LWG-3571 flush_emit should set badbit if the emit call fails #2396)
  • LWG-3574 common_iterator should be completely constexpr-able (filed LWG-3574 common_iterator should be completely constexpr-able #2397)
  • LWG-3580 iota_view's iterator's binary operator+ should be improved (filed LWG-3580 iota_view's iterator's binary operator+ should be improved #2398)
  • LWG-3589 The const lvalue reference overload of get for subrange does not constrain I to be copyable when N == 0 (filed LWG-3589 The const lvalue reference overload of get for subrange does not constrain I to be copyable when N == 0 #2399)
  • LWG-3590 split_view::base() const & is overconstrained (filed LWG-3590 split_view::base() const & is overconstrained #2400)
  • LWG-3591 lazy_split_view<input_view>::inner-iterator::base() && invalidates outer iterators (filed LWG-3591 lazy_split_view<input_view>::inner-iterator::base() && invalidates outer iterators #2401)
  • LWG-3592 lazy_split_view needs to check the simpleness of Pattern (filed LWG-3592 lazy_split_view needs to check the simpleness of Pattern #2402)
  • LWG-3593 Several iterators' base() const & and lazy_split_view::outer-iterator::value_type::end() missing noexcept (filed LWG-3593 Several iterators' base() const & and lazy_split_view::outer-iterator::value_type::end() missing noexcept #2403)
  • LWG-3595 Exposition-only classes proxy and postfix-proxy for common_iterator should be fully constexpr (filed LWG-3595 Exposition-only classes proxy and postfix-proxy for common_iterator should be fully constexpr #2264)
• PR out for review

[MattStephanson]

LWG-3561: This would involving changing discard_block's internal _Nx counter and template arguments from int to size_t, which I think would be an ABI break. I think the most we could do now is add a // TRANSITION, ABI comment, and maybe a static_assert for discard_block_engine that its parameters must be <= INT_MAX.

[CaseyCarter]

LWG-3561: This would involving changing discard_block's internal _Nx counter and template arguments from int to size_t, which I think would be an ABI break. I think the most we could do now is add a // TRANSITION, ABI comment, and maybe a static_assert for discard_block_engine that its parameters must be <= INT_MAX.

I think we could greatly improve discard_block_engine by divorcing it completely from the TR1-era discard_block. We could even make it tolerate p > INT_MAX, but I agree that r > INT_MAX is not supportable without breaking ABI.

[timsong-cpp]

LWG-3561: This would involving changing discard_block's internal _Nx counter and template arguments from int to size_t, which I think would be an ABI break. I think the most we could do now is add a // TRANSITION, ABI comment, and maybe a static_assert for discard_block_engine that its parameters must be <= INT_MAX.

I think we could greatly improve discard_block_engine by divorcing it completely from the TR1-era discard_block. We could even make it tolerate p > INT_MAX, but I agree that r > INT_MAX is not supportable without breaking ABI.

Couldn't you change the type of _Nx only if r > INT_MAX?

Edit: looks like MSVC doesn't check narrowing from size_t to int in the NTTP, hmm...is it an ABI break if the existing thing compiles but totally doesn't work?

[MattStephanson]

Couldn't you change the type of _Nx only if r > INT_MAX?

Let me create a new issue for LWG-3561 so that this meta-issue doesn't become too crowded

[AlexGuteniev]

• LWG-2762 unique_ptr operator*() should be noexcept - filed LWG-2762 unique_ptr operator*() should be noexcept #2257, good-first-issue candidate. It is also a bug as the current noexcept for unique_ptr is too strengthened.

[cpplearner]

Following issues are Already implemented:

LWG-3568 basic_istream_view needs to initialize value_

STL/stl/inc/ranges

Line 1355 in dc888f7

_Ty _Val = _Ty{}; // Per LWG issue submitted but unnumbered as of 2021-06-15

LWG-3535 join_view::iterator::iterator_category and ::iterator_concept lie

STL/stl/inc/ranges

Line 2967 in dc888f7

conditional_t<common_range<_Inner> // per LWG-3535

STL/stl/inc/ranges

Line 3039 in dc888f7

// Per LWG issue unnumbered as of 2021-03-16

LWG-3498 Inconsistent noexcept-specifiers for basic_syncbuf

STL/stl/inc/syncstream

Line 109 in dc888f7

basic_syncbuf& operator=(basic_syncbuf&& _Right) { // see LWG-3498 regarding noexcept

STL/stl/inc/syncstream

Line 117 in dc888f7

void swap(basic_syncbuf& _Right) { // see LWG-3498 regarding noexcept

LWG-3581 The range constructor makes basic_string_view not trivially move constructible

STL/stl/inc/xstring

Line 1287 in dc888f7

requires (!same_as<remove_cvref_t<_Range>, basic_string_view> // per LWG-3581

LWG-3293 move_iterator operator+() has incorrect constraints

STL/stl/inc/xutility

Line 3865 in dc888f7

{ _Right.base() + _Off } -> same_as<_Iter>; // LWG-3293

[AlexGuteniev]

• LWG-3595 Exposition-only classes proxy and postfix-proxy for common_iterator should be fully constexpr - LWG-3595 Exposition-only classes proxy and postfix-proxy for common_iterator should be fully constexpr #2264

[cpplearner]

These seem to be Already implemented (some might be closer to the "Not applicable" side, e.g. LWG-3559):

LWG-2191 Incorrect specification of match_results(match_results&&)

LWG-2381 Inconsistency in parsing floating-point numbers

LWG-3123 duration constructor from representation shouldn't be effectively non-throwing

LWG-3152 common_type and common_reference have flaws in common

LWG-3361 safe_range<SomeRange&> case

LWG-3559 Semantic requirements of sized_range are circular

LWG-3560 ranges::equal and ranges::is_permutation should short-circuit for sized_ranges

LWG-3567 Formatting move-only iterators, take two

LWG-3572 copyable-box should be fully constexpr

LWG-3573 Missing Throws element for basic_string_view(It begin, End end)

LWG-3585 variant converting assignment with immovable alternative

LWG-3407 Some problems with the wording changes of P1739R4

STL/stl/inc/ranges

Line 2437 in 8833270

// The following are all per the proposed resolution of LWG-3407

STL/stl/inc/ranges

Line 2807 in 8833270

// The following are all per the proposed resolution of LWG-3407

[StephanTLavavej]

@cpplearner Thanks! I've updated the list accordingly with the results of your analysis. The one exception was LWG-2381 - I noticed it still needs implementation work, so I filed a tracking issue with a repro.