ISO standards macro are rejected by g++

[Marc Espie]

In article <flbt2yyn6i.fsf@sel.cmla.ens-cachan.fr> you write:
>Zack Weinberg <zack@wolery.cumb.org> writes:

>More generally, g++ does not (currently) reject some of GNU extenion
>identifiers beginning with __.  I'm not sure what should be the right 
>approach. For example I don't want g++ to complain about __complex__
>in the standard header <complex>.  I guess one could complain
>conditionally according to some flags  (for example testing 
>in_system_header_p). 

Normally, gnu extensions necessary for standard headers are flagged
with __extension__, which is explicitly designed to get pedantic to shut
up in that case.