[Patch] libstdc++/25626: valarray vs non-POD

[Paolo Carlini]

Paolo Carlini wrote:

>>The gslice_array bits are not ok: _M_data is private. I'm looking into this. 
>>    
>>
>I can deal with this problem taking another small bit from Gaby's big draft.
>  
>

However, I like the below much more (*), even if it adds a little bit of
code and copy_construct goes through an indirection: is *totally*
consistent with the existing design.

Tested x86-linux + a few non-GPL tests.

Paolo.

(*) In fact it's simply an extended version of my original sketch:

    http://gcc.gnu.org/ml/libstdc++/2005-12/msg00084.html

//////////////
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: CL_25626_3
URL: <http://gcc.gnu.org/pipermail/libstdc++/attachments/20060114/ad5de2eb/attachment.ksh>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: patch_25626_3
URL: <http://gcc.gnu.org/pipermail/libstdc++/attachments/20060114/ad5de2eb/attachment-0001.ksh>