Subject: Re: gcc3 patches
To: Johnny C. Lam <jlam@netbsd.org>
From: Manuel Bouyer <bouyer@antioche.lip6.fr>
List: tech-pkg
Date: 02/24/2004 13:45:59
On Tue, Feb 24, 2004 at 09:28:45AM +0000, Johnny C. Lam wrote:
> On Mon, Feb 23, 2004 at 07:02:21PM +0100, Manuel Bouyer wrote:
> > On Mon, Feb 23, 2004 at 06:57:06PM +0100, Manuel Bouyer wrote:
> > > Hi,
> > > here are various patches against the current gcc3 packages and dependancy,
> > 
> > This time with the patch (thanks wiz)
> > 
> > > which do various things:
> > > - upgrade to gcc 3.3.3 (the most important, 3.3.2 can't build qt3-tools
> > >   on solaris because of what looks like a codegen bug)
> > > - remove /usr/local/{include,lib} from the default search path, because this
> > >   breaks a lot of things when LOCALBASE=/usr/local
> > > - handle non-empty ${GCC_VERSION} better (but in its current form breaks with
> > >   an empty ${GCC_VERSION}).
> > > 
> > > Comments ?
> 
> I don't understand the purpose of patch-ah.  Could you please explain
> what that's trying to do?
> It looks like it's just removing some
> dependencies for config.status, but why is that necessary?  Also the

Sorry, I forgot to comment on this one.
On my system, $(srcdir)/configure timestamp keep being updated, which cause
a loop.
I just noticed there is:
	cd $(srcdir) && $(AUTOCONF)
just after config.status run, which may be the real cause of the problem,
it may be worth trying to remove just this line, instead of the dependancy.

Note that I'm running a very recent version of solaris9 (12/03 build).


> changes that add ${GCC3_INSTALLTO_PREFIX} everywhere are incorrect in
> the default case where that value is empty as it leaves an underscore
> in the name of the various gcc3-* packages.

Yes. It was correct for the (short) period of time where GCC3_INSTALLTO_PREFIX
was never empty :)
As I said, this part needs to be fixed.
${GCC3_PKGMODIF} should probably be used for this.  The problem I have with it
is that it's in Makefile.common, so it's not available to buildlink or
compilers/*.mk

Should the GCC3_PKGMODIF code be duplicated, or is there a central place where
it could be ?

--
Manuel Bouyer, LIP6, Universite Paris VI.           Manuel.Bouyer@lip6.fr
     NetBSD: 24 ans d'experience feront toujours la difference
--