From: John Maddock (jm_at_[hidden])
Date: 2003-04-16 07:35:42
> Sure. It's actually a patch for is_enum.hpp but it fixes warnings for
> both is_class and is_enum, namely:
> test::is_class< int( int ) >::value
> test::is_enum< int( int ) >::value
> I added the fix globally, but if you prefer to remove the dependency to
> is_function for other compilers, you can add a guard.
Yes, I made it gcc specific, also added a similar fix for is_union, but not
for has_trivial_copy which also produces these warnings. Fixes will be in
cvs towards the weekend - when I get my cvs access up again (I only have
limited mail access at present).
> Also, IIRC, the comment from Dave seen below is lying (due to changes
> that happened after he wrote it). Maybe we should fix this up, too. It
> would IMHO lead to the removal of the #ifdef-block, but I didn't wanted
> to mix it up with the fix for the warnings, so I left it out for now.
Not quite - the fix is required if you can't pass non-copyable types through
is_convertible, and that is still true for some compilers unfortunately,
even if this has been fixed for most of them...