[Patch/RFC] libstdc++/19343

[Paolo Carlini]

Gabriel Dos Reis wrote:

>Yes.  The way Andreas analysis came to me was that he wanted to matke
>the _M_get() function noreturn.
>  
>
Ok, fine.

By the way, I'm committing a version of the patch which uses:

    __attribute((__noreturn__))

instead of

    __attribute__((noreturn))

as suggested in the audit trail. Indeed, seems more correct to me wrt 
namespace
pollution. Actually, I think we should audit the whole library wrt this 
issue. Do
you agree?

Paolo.