vfwscanf() and friends

[Hasso Tepper]

Simon 'corecode' Schubert wrote:
> Hasso Tepper wrote:
> > Looking at headers it seems to me that libstdc++ assumes that if
> > vfwscanf() and friends are in libc, they are there unconditionally
> > which isn't the case - they are visible in /usr/include/wchar.h if
> > source defines _POSIX_C_SOURCE at least 200112 (C99).
> 
> I think we can declare them in any case.  After all, we're all moving
> to C99, right? 

This isn't probably good idea. After all I still want to develop programs
which would use only Cxx features where xx < 99 using DragonFlyBSD. Fixing
system compiler would be acceptable.

> > So, question - how to fix it and where? In libstdc++? But this will
> > still cause problems to anyone not using compiler provided by system.
> > In our /usr/include/wchar.h?
> 
> Where do other systems handle this?  Using a non-system compiler is
> not feasible to support anyways ... 

There is comment #1 in http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24012
which explains how it's dealt in Linux - _GNU_SOURCE is defined all the
time.

Anyway, at the moment I'm thinking that unless we just fix the bug in
compiler, we shouldn't do anything. Just deal with the fact that all
current GCC versions are buggy in the code, not in the system.

> Anybody wanna add thread support to tinycc?  

:)


regards,

-- 
Hasso Tepper
Elion Enterprises Ltd. [AS3249]
IP & Data Networking Expert