On 21/02/22 21:54, Jonathan Wakely wrote:
On Mon, 21 Feb 2022 at 18:00, François Dumont via Libstdc++ <libstd...@gcc.gnu.org <mailto:libstdc%2b...@gcc.gnu.org>> wrote:Gentle reminder, it is important to have this for gcc 12.Well, it's been broken since 4.8, so another year wouldn't be the end of the world ;-)
Sorry for the pressure, I thought I had broken it with my fix of PR 96088. Which moreover was in gcc 11.
So indeed not mandatory for gcc 12 but still nice to fix eventually. Thanks
I did start reviewing it, but I was trying to find a simpler way to solve it than adding all those overloads. I'll take another look tomorrow and either approve your patch or suggest something else.On 15/02/22 10:05, François Dumont wrote: > We have a regression regarding management of types convertible to > value_type. It is an occurrence of PR 56112 but for the insert method. > > libstdc++: [_Hashtable] Insert range of types convertible to > value_type PR 56112 > > Fix insertion of range of types convertible to value_type. > > libstdc++-v3/ChangeLog: > > PR libstdc++/56112 > * include/bits/hashtable.h > (_Hashtable<>::_M_insert_unique_aux): New. > (_Hashtable<>::_S_to_value): New. > (_Hashtable<>::_M_insert(_Arg&&, const _NodeGenerator&, > true_type)): Use latters. > * testsuite/23_containers/unordered_map/cons/56112.cc: Use > dg-do compile. > * testsuite/23_containers/unordered_set/cons/56112.cc: New > test. > > Tested under Linux x86_64. > > Ok to commit ? > > François
