Re: cpplib: Dump __GXX_WEAK__ and __STDC__ with -dM

[Rainer Orth <ro at TechFak dot Uni-Bielefeld dot DE>]

> IMHO, the best test would be to bootstrap with and without the patch
> and see exactly what the header file diffs in gcc/include are.  Once
> we see that, then we'll know if there's a problem.  I suspect the
> original issues prompting the hack remains.

That's what I did, and there are no differences between those directories.

> IIRC, STDC_0_IN_SYSTEM_HEADERS was introduced to address a specific
> problem, namely that we *don't* want fixincludes to diddle every
> header file on solaris2.  There are numerous reasons to want to lower
> the number of files gratuitously changed, I won't rehash them here.

But this is not what happens at the moment: we both run fixincludes to fix
that __STDC__-0==0 stuff and still have STDC_0_IN_SYSTEM_HEADERS, which
won't trigger any longer since we have fixed copies of all affected system
headers anyway.  This is a change from gcc 2.95.2 where fixincludes wasn't
run on Solaris 2.5 and up and thus STDC_0_IN_SYSTEM_HEADERS was necessary
for gcc to work with the unchanged vendor headers; I haven't check when
this was introduced.

	Rainer

-----------------------------------------------------------------------------
Rainer Orth, Faculty of Technology, Bielefeld University

Email: ro@TechFak.Uni-Bielefeld.DE