[v3] First bits of libstdc++/44436

[Paolo Carlini]

On 10/28/2010 06:11 PM, Paolo Carlini wrote:
> However, I'd like to ask your help (I'm addressing in particular the
> library people of course) in figuring out a long term solution for the
> _Select1st function object: the one we traditionally use in C++03 mode
> doesn't work with move-semantics, because insists on using
> container::value_type (the -2 testcases would not compile with it). For
> the associative containers I guess the situation will be completely
> similar and I'd like to figure out a satisfactory solution looking
> forward, possibly improving the std::_Select1st itself and just using it
> everywhere. Is __detail::_Select1st already ok for C++0x mode or we want
> something better?
>   
About this issue, I'm committing the below, a definite improvement,
IMHO. With it, the associative containers should also be ok.

Tested x86_64-linux

Paolo.

/////////////////////////////
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: CL_select1st
URL: <http://gcc.gnu.org/pipermail/libstdc++/attachments/20101029/fea9dbb0/attachment.ksh>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: patch_select1st
URL: <http://gcc.gnu.org/pipermail/libstdc++/attachments/20101029/fea9dbb0/attachment-0001.ksh>