This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: compilation of libio failing on linux/libc6
- To: jhpb at sarto dot gaithersburg dot md dot us
- Subject: Re: compilation of libio failing on linux/libc6
- From: Manfred dot Hollstein at ks dot sel dot alcatel dot de
- Date: Tue, 27 Jan 98 16:04:54 +0100
- Cc: egcs at cygnus dot com, hjl at lucon dot org, drepper at cygnus dot com
- References: <9801270815.AA08684@lts.sel.alcatel.de> <199801271243.HAA14935@altera.gaithersburg.md.us>
- Reply-To: Manfred dot Hollstein at ks dot sel dot alcatel dot de
On Tue, 27 January 1998, 07:43:40, jhpb@sarto.gaithersburg.md.us wrote:
> >>>>> "MH" == Manfred Hollstein <Manfred.Hollstein@ks.sel.alcatel.de> writes:
>
> MH> I guess this could be caused by the latest change to the top-level configure
> MH> script which was emitting the various Makefile fragments in the wrong order.
>
> MH> Could you please post the contents of `libraries/libio/Makefile'?
>
> Here you go.
>
[Makefile deleted]
Yup, I just saw in the top-level ChangeLog:
1997-09-15 02:37 Ulrich Drepper <drepper@cygnus.com>
* configure: Rewrite so that project Makefile fragment is inserted
first and appears last in the resulting Makefile.
But this is contrary to the comment in `configure':
# the four makefile fragments MUST end up in the resulting Makefile in this order:
# package, target, host, and site.
And I still believe that this order is correct; to fix the Linux stuff I think
we need to split the package_fragment_stuff into two parts: one defining package
related macros and one defining rules, dependancies, etc.
The final Makefile then will contain in this order:
1. srcdir, VPATH, other configure'd macros
2. package macros
3. target
4. host
5. site
6. package rules
7. ...
In 2. package macros we then define the 99% default:
_G_CONFIG_H=_G_config.h
while in 3. linux.mt we'll override that by
_G_CONFIG_H=
and _before_ it's value will be used in any rule, ...
I'll send a patch for this.
Manfred