This is the mail archive of the
libc-hacker@sourceware.cygnus.com
mailing list for the glibc project.
Re: Problems installing glibc in /usr/local
- To: Zack Weinberg <zack@rabi.phys.columbia.edu>
- Subject: Re: Problems installing glibc in /usr/local
- From: Roland McGrath <roland@frob.com>
- Date: Wed, 9 Sep 1998 11:55:05 -0400
- Cc: Andreas Jaeger <jaeger@informatik.uni-kl.de>, libc-hacker@cygnus.com
- Emacs: you'll understand when you're older, dear.
> I think we should override gcc's <limits.h> entirely.
Those definitions should not be in maintained two separate places in two GNU
packages.
> This would be easy -
> just take out the #include_next and the #ifndef around the ANSI limits
> definitions. We'd have to specialize those limits to each processor, but
> that should only require a bits/ansi_lim.h for wordsize-32 and wordsize-64.
> gcc's header already defers to the one we install.
That misstates the intent of the existing code. gcc's header uses
#include_next in the way it does (with various #ifdef's) precisely so that
glibc's header and gcc's header will cooperate as they do now to have gcc
define all the macros about C types, and have glibc define all the macros
about POSIX stuff. That code was not put in gcc's limits.h "to defer to"
libc's; it was put there to cooperate with glibc's limits.h in the specific
way that they do now, and also was hoped to perhaps cooperate with another
system's limits.h or lack thereof.