This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: where are char_traits<wchar_t> instances built ?
- From: Marc Espie <espie at nerim dot net>
- To: Paolo Carlini <pcarlini at suse dot de>
- Cc: gcc at gcc dot gnu dot org
- Date: Fri, 2 Jan 2004 17:49:19 +0100
- Subject: Re: where are char_traits<wchar_t> instances built ?
- References: <20031229140434.GA23567@tetto.gentiane.org> <3FF0370E.5000307@suse.de>
- Reply-to: espie at nerim dot net
On Mon, Dec 29, 2003 at 03:15:42PM +0100, Paolo Carlini wrote:
> Marc Espie wrote:
>
> >Looking in my libstdc++, I see no instance of
> >std::char_traits<T>::compare, either for T=char nor T=wchar.
>
> Hummm. Could you provide additional information on the actual error?
>
> Usually, ../include/c++/3.3.2/bits/char_traits.h contains struct
> char_traits and its specializations for char and wchar_t, quite
> straightforward, by the way (in 3.3.x, at least): as you can easily
> imagine they boil down to just memcmp, memchr, memmove and so on...
>
The #ifdef _GLIBCPP_USE_WCHAR_T is disabled, as the system couldn't
find the required functions.
On the other hand, gcc 2.95 did a have a generic set of functions which
did at least allow code with wchar_t to compile, even though the result
might not have been that useful...
This is a definite regression. Not a big one, but still a regression for
me. id3lib used to compile like a charm, but no longer...