[libc++] __int128 support in type_traits

libc++'s type_traits header doesn’t properly support |__int128 and
__uint128 integers as type parameter values for is_integral, is_signed,
is_unsigned, make_signed and make_unsigned. Is this by design, or would
a patch be welcome?

I think this would be great personally.

A related question is about long double.

On Windows, what to do about sizeof(long double) for MSVC being 8 and sizeof(long double) for g++ being 16.

G M <gmisocpp@gmail.com> writes:

On Windows, what to do about sizeof(long double) for MSVC being 8 and
sizeof(long double) for g++ being 16.

s/16/12