Meta LWG issue: 2021-06 meeting

[StephanTLavavej]

(Previous meta-issue: #1679)

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

❔ Not yet analyzed

• All done! 😺

For each of these issues, we've determined whether they are:

✖️ 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-2997 LWG-491 and the specification of {forward_,}list::unique
  • LWG-3521 Overly strict requirements on qsort and bsearch
  • LWG-3526 Return types of uses_allocator_construction_args unspecified
• 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
  • LWG-2818 "::std::" everywhere rule needs tweaking
  • LWG-3462 Formatter requirements forbid use of fc.arg()
  • LWG-3551 borrowed_{iterator,subrange}_t are overspecified

😸 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

  • LWG-3481 viewable_range mishandles lvalue move-only views (Removal in <ranges>: Remove "per LWG" comments #1967)
  • LWG-3517 join_view::iterator's iter_swap is underconstrained (Removal in <ranges>: Remove "per LWG" comments #1967)
  • LWG-3523 iota_view::sentinel is not always iota_view's sentinel (Removal in <ranges>: Remove "per LWG" comments #1967)
  • LWG-3530 BUILTIN-PTR-MEOW should not opt the type out of syntactic checks (Removal in cleared comment TRANSITION  #2186)
  • LWG-3532 split_view<V, P>::inner-iterator<true>::operator++(int) should depend on Base (Removal in <ranges>: Remove "per LWG" comments #1967)
  • LWG-3543 Definition of when counted_iterators refer to the same sequence isn't quite right (Filed as LWG-3543 Definition of when counted_iterators refer to the same sequence isn't quite right #2184)

• Implemented without comments

  • LWG-3410 lexicographical_compare_three_way is overspecified
  • LWG-3430 std::fstream & co. should be constructible from string_view (Implemented in <fstream>: Implement LWG-3430 #1968)
  • LWG-3506 Missing allocator-extended constructors for priority_queue (Implemented in Implement P1425 iterator range constructors for stack and queue #1994)
  • LWG-3518 Exception requirements on char trait operations unclear (Implemented in Use LWG-3518 to add noexcept #2080)
  • LWG-3522 Missing requirement on InputIterator template parameter for priority_queue constructors (Implemented in Implement P1425 iterator range constructors for stack and queue #1994)
  • LWG-3527 uses_allocator_construction_args handles rvalue pairs of rvalue references incorrectly (Implemented in implement LWG-3527 #2194)
  • LWG-3528 make_from_tuple can perform (the equivalent of) a C-style cast (Implemented in <tuple>: LWG-3528 make_from_tuple can perform (the equivalent of) a C-style cast #2160)
  • LWG-3529 priority_queue(first, last) should construct c with (first, last)
  • LWG-3533 Make base() const & consistent across iterator wrappers that support input_iterators (Implemented in Adopt LWG-3533 Make base() const & consistent across iterator wrappers that support input_iterators #1993)
  • LWG-3536 Should chrono::from_stream() assign zero to duration for failure? (Implemented in <chrono> LWG-3536: leave duration unchanged on failed parse #2037)
  • LWG-3539 format_to must not copy models of output_iterator<const charT&>
  • LWG-3540 There should be no const in basic_format_arg(const T* p)
  • LWG-3541 indirectly_readable_traits should be SFINAE-friendly for all types (Implemented in Adopt LWG-3541 indirectly_readable_traits should be SFINAE-friendly for all types #1992)
  • LWG-3542 basic_format_arg mis-handles basic_string_view with custom traits
  • LWG-3544 format-arg-store::args is unintentionally not exposition-only
  • LWG-3548 shared_ptr construction from unique_ptr should move (not copy) the deleter (Implemented in Implement LWG-3548 shared_ptr construction from unique_ptr should move (not copy) the deleter #2007)
  • LWG-3553 Useless constraint in split_view::outer-iterator::value_type::begin() (Implemented in Adopt LWG-3553 Useless constraint in split_view::outer-iterator::value_type::begin() #1990)
  • LWG-3555 {transform,elements}_view::iterator::iterator_concept should consider const-qualification of the underlying range (Implemented in Adopt LWG-3555 {transform,elements}_view::iterator::iterator_concept should consider const-qualification of the underlying range #1989)

🩹 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.

• LWG-3552 Parallel specialized memory algorithms should require forward iterators (Mentioned in <memory>: Missing C++17 ExecutionPolicy overloads #525)

🐞 Not yet implemented

• PRs out for review:
  • LWG-2774 std::function construction vs assignment (Implemented in LWG-2774 std::function construction vs assignment #2098)
  • LWG-3546 common_iterator's postfix-proxy is not quite right (Implemented in Adopt LWG-3546 common_iterator's postfix-proxy is not quite right #1991)
• We should file a GitHub issue labeled LWG; see those existing examples.
  • LWG-3519 Incomplete synopses for <random> classes (Filed as LWG-3519 Incomplete synopses for <random> classes #2179)
  • LWG-3520 iter_move and iter_swap are inconsistent for transform_view::iterator (Filed as LWG-3520 iter_move and iter_swap are inconsistent for transform_view::iterator #2180)
  • LWG-3549 view_interface is overspecified to derive from view_base (Filed as LWG-3549 view_interface is overspecified to derive from view_base #2178)

[SuperWig]

The only LWG issue cited in the code base, other than the ones mentioned in that PR I just submitted, is LWG-3536 which isn't implemented.

[cpplearner]

LWG-3539 format_to must not copy models of output_iterator<const charT&>

Already implemented in #1795. (It seems that the proposed resolution in LWG-3539 only touches format_to, but as shown in #1795, format_context also needs to be changed to not copy the iterator.)

LWG-3540 There should be no const in basic_format_arg(const T* p)

Already implemented in #1867

[CaseyCarter]

LWG-2774 is not yet implemented.
LWG-2818 is a pure wording change; there's nothing to implement.
LWG-2997 is a wording cleanup; there's nothing to implement.
LWG-3410 is a relaxation that we already take advantage of (call this one "implemented").
LWG-3430 is not yet implemented (but #1968 implements it).
LWG-3462 is a relaxation that we already take advantage of (ditto "implemented").

@SuperWig's #1967 removes comments for the already-implemented:

• LWG-3481
• LWG-3523
• LWG-3517
• LWG-3532

LWG-3539 format_to must not copy models of output_iterator<const charT&>

Already implemented in #1795. (It seems that the proposed resolution in LWG-3539 only touches format_to, but as shown in #1795, format_context also needs to be changed to not copy the iterator.)

I managed to bungle the issue submission; I've submitted a followup issue with the basic_format_context changes.

[MattStephanson]

LWG-3536 is tracked (and was prompted) by #1740.

[miscco]

LWG-3552 does not apply as there are currently no parallel specialized memory algorithms

[miscco]

Created a PR for LWG-3555 here #1989

[miscco]

Regarding LWG-3551 I do not see any necessary changes. As far as I remember MSVC internally optimizes conditional_t away and we do not have an alternative mechanism at hand