This is the mail archive of the
egcs-patches@egcs.cygnus.com
mailing list for the EGCS project.
Re: Your g++ breaks glibc.
- To: mark@codesourcery.com
- Subject: Re: Your g++ breaks glibc.
- From: Bruce Korb <autogen@orion.linuxbox.com>
- Date: Thu, 15 Jul 1999 12:53:58 -0400 (EDT)
- Cc: hjl@varesearch.com, jason@cygnus.com, egcs-patches@egcs.cygnus.com, libc-hacker@sourceware.cygnus.com, bje@cygnus.com, korb@datadesign.com
> And I'll defer to Jason. Personally, I'd like to see the code that is
> extra generous with system headers yanked; vendors should fix their
> system headers, or, if necessary, we should fixinclude them.
^^^^^^^^^^
I found this keyword :-)
but I did not find enough context in egcs-patches to be sure
I know where the bogon header problem lies. It seems to be a bit
of a "fixheaders" problem (which probably needs folding into
fixincludes anyway...).
> it may not know which one,
>
> extern "C" void exit(int);
> extern "C" void exit(int) throw ();
>
> to use for a given system. gcc 2.95 works with
>
> #include <stdlib.h>
> extern "C" void exit(int);
>
> but not
>
> extern "C" void exit(int);
> #include <stdlib.h>
>
> I don't see any reason to allow one and not the other.
It seems reasonable, but then standards do not always follow
"reason" :-(. Since I don't have a standard handy,
which one is supposed to be allowable and are we supposed
to be fixing stdlib.h or cstdlib?
Regards,
Bruce