Re: [v3] Fix / clean-up config vs crosses (2/n)

[Mark Mitchell <mark at codesourcery dot com>]

Paolo Carlini wrote:
> Yeah, now I remember why Joseph introduced AC_TRY_COMPILE_OR_LINK, for 
> some C99 functions... I think we already said that the practice isn't 
> really nice, for users too, not just for our configury... Anyway, can I 
> ask to you and Joseph to double check whether those 
> declared-but-not-defined tricks are common also for functions like 
> writev, pool, LFS I/O... Certainly you know those systems much better 
> than me...

Actually, I don't know much about them, except from a very high-level 
perspective.  I could go dive into it, of course, but for me this is 
more theory than practice, I'm afraid.  There are so many systems out 
there that I just think it pays to be paranoid.

Concretely, I have no evidence of a system which puts writev in headers 
without guaranteeing it appears in the C library, but I can imagine it.

Even with Newlib, for example, people building freestanding systems 
leave out libgloss (or requivalent) -- but the headers don't change. 
So, the entire I/O subsystem (open, read, write, etc.) might be missing. 
 On the other hand, in such a situation, there's still nothing wrong 
with having libstdc++ I/O (cout, etc.) use the libc I/O routines; if the 
end user fills in the low-level I/O routines, then libstdc++ will work, 
and if they don't, then libstdc++ I/O isn't going to work no matter 
what.  So, the issue is really about "optional" I/O routines that 
libstdc++ could live without.

Thanks,

--
Mark Mitchell
CodeSourcery
mark@codesourcery.com
(650) 331-3385 x713