[Patch] libstdc++/21244

[Paolo Carlini]

Paolo Carlini wrote:

>Ok, therefore, let's not name anything (I have to revert the fix for 21244).
>
>Then, which cure are you suggesting? Shall I add casts to an
>"appropriate" type everywhere? If this is the only real solution, let's
>go with it. Probably, besides the case of cpp_type_traits, in most the
>other places int would be ok. Certainly, when complex arithmetic is
>involved we are going to uglify a lot the code, sigh.
>  
>
I went through the various occurrences and maybe the situation is not
really horrible: in most of the places, minimal arithmetic is carried
out. In other places the cast is either already present (e.g., case of
stl_deque.h) or the right type to use is explicit (e.g., case of
formatter.h). I wil prepare a complete patch and post it, ASAP.

Thanks again for your feedback,
Paolo.