https://gcc.gnu.org/bugzilla/show_bug.cgi?id=122313
Jonathan Wakely <redi at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
See Also| |https://github.com/microsof
| |t/STL/issues/107
Keywords| |wrong-code
--- Comment #12 from Jonathan Wakely <redi at gcc dot gnu.org> ---
(In reply to Jiang An from comment #10)
> I'm not very sure. My LLVM PR 116001 only touched vector<nonbool>.
> vector<bool> was left untouched due to
> https://cplusplus.github.io/LWG/issue3297.
Until LWG accepts that issue, it seems that the current wording makes the
original testcase in this bug report invalid. Although we should do an explicit
construction using allocator_traits::construct, the iterator value type *also*
needs to be implicitly convertible (even though that's not actually supposed to
be used).
Libstdc++ still has a bug, because the test4() case using pmr::vector shows
that we do the wrong thing even when the implicit convertibility requirement is
met. So we should fix this anyway, and if/when LWG accepts issue 3297 we'll be
ready..
