This is the mail archive of the
newlib@sourceware.org
mailing list for the newlib project.
Re: [PATCH/RFA] Internationalize ctype functionality
On Apr 1 17:13, Corinna Vinschen wrote:
> On Apr 1 15:46, Hans-Peter Nilsson wrote:
> > That part was already there, effectively... oh, you mean some
> > deliberate means; testing _MB_CAPABLE. Right. Patch
> > forthcoming. After *cough* testing (though it'll just be
> > cris-elf, a !_MB_CAPABLE target; not wanting to...raise the
> > testing bar).
>
> As I wrote in my previous mail, I have included that in my local version
> of the patch adding the missing code for the !ALLOW_NEGATIVE_CTYPE_INDEX
> case. I've included the entire patch agin in this mail, for completeness.
>
> I built ctype_.c with excessive warning level for all combinations of
> ALLOW_NEGATIVE_CTYPE_INDEX and _MB_CAPABLE and gcc seems to be happy now.
>
>
> Corinna
>
>
> * libc/ctype/ctype_c.c: Move inclusion of ctype_iso.h and
> ctype_cp.h out of ALLOW_NEGATIVE_CTYPE_INDEX case.
> (_ctype_b): Constify in !_MB_CAPABLE case. Restrict !static
> to Cygwin.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
This last part about only constifying _ctype_b in the
!_MB_CAPABLE case was a bit over the top. It works but is
nonsense. Except for Cygwin, which needs write access for
backward compatibility, all other targets are quite happy
to have _ctype_b `static const'. I reverted this snippet
of the patch locally.
Corinna
--
Corinna Vinschen
Cygwin Project Co-Leader
Red Hat