svn commit: trunk/uClibc: include

Denys Vlasenko vda.linux at googlemail.com
Tue Jan 13 11:26:32 UTC 2009


On Tuesday 13 January 2009 11:48, Bernd Schmidt wrote:
> Matthieu CASTET wrote:
> > But the external ABI shouldn't depend of signedness of char : this will
> > make impossible to do portable application.
> > 
> > Do you got some example where char is used in the ABI and sign is needed ?
> 
> In the old thread I pointed out that a macro in a header files may have
> two meanings depending on whether it's included by uClibc or by
> applications.  It's just not something we should risk for dubious gain.

The goal should be that macro works correctly regardless
of char signedness. Reasons:

* We do not want to have different macros for different
  architectures. I don't think adding more to the ifdef forest
  in ctype.h is appealing to anyone. So, ctype.h macros
  won't be forked for "signed char" architentures
  and for "unsigned char" architentures. Therefore they
  should be agnostic to that.

* We want users to be able to use -funsigned-char or
  -fsigned-char when they build their programs against
  uclibc, and still have their programs working correctly.

--
vda


More information about the uClibc mailing list