> Should we really be ignoring this? I would have thought we should look for
> the last of -fsigned-char, -fno-unsigned-char, -funsigned-char,
> -fno-signed-char, and pick a signed type unless the last flag (if any) is
> one of the latter two.
>
> For instance, -funsigned-char -fno-unsigned-char should probably leave char
> signed.
>

That is what gcc does. With the command

gcc -Wall -std=c++11 test.cpp -S -funsigned-char -fno-unsigned-char

gcc compiles

#include <type_traits>
static_assert(std::is_signed<char>::value, "foo");

Cheers,
Rafael
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

Reply via email to