[SM-Discuss] glibc - test grimoire

Arwed von Merkatz v.merkatz at gmx.net
Tue Feb 22 15:05:41 EST 2005


On Tue, Feb 22, 2005 at 11:56:48AM -0800, Seth Alan Woolley wrote:
> On Tue, Feb 22, 2005 at 08:34:12PM +0100, Arwed von Merkatz wrote:
> > On Tue, Feb 22, 2005 at 09:53:00AM -0800, Eric Sandall wrote:
> > > Quoting Seth Alan Woolley <seth at positivism.org>:
> > > <snip>
> > > > Arwed, Eric, shall I do this?  How long after I do this do we bump gcc
> > > > 3.4.x to test?
> > > 
> > > I'd say glibc is just about ready for stable if we don't receive any more
> > > reports of breakage by tomorrow (February 23, 2005).
> > 
> > Sounds good.
> 
> Tomorrow in your time?  That's 6 PM here. ;)

Tomorrow in my time is in three hours ;)

> > 
> > > As for gcc, I believe it's ready to go as well except for it breaking the
> > > compilation of openoffice[0]. Arwed?
> > 
> > I was thinking of waiting until glibc hits the stable grimoire before
> > getting gcc to test so the stable users don't have two major upgrades at
> > the same time. Not sure what to do with it breaking openoffice. How many
> > people are actually crazy enough to compile that beast from source?
> > Other than that gcc 3.4 can go in as soon as the glibc update in stable
> > is done.
> 
> It might be worth a gcc3.3 spell like we had with USEGCC2 a while back.  
> It might be smart to allow us to have many versions of gcc installed and 
> instead of USEGCCn have a DETAILS:GCC=3.3 variable.  I'm pretty sure gcc 
> problems will happen again in the future during upgrades, so it just 
> might be worth it.
> 
> It might... might.. might. might mightmightightghthtt...
> 
> Anyways, you get the point.

I just had a different idea, though I don't know if it works:
Create a gcc 3.3 spell that installs to /usr as usual but uses the
suffix '3.3' for the binaries. All stuff gcc installs is versioned
except for the binaries, so if it does have a configure switch to add a
suffix to binaries (usually autoconf build systems do), that could work.
Then there wouldn't have to be any support for it in sorcery, spells
needing it would just depend on it and set CC=gcc3.3 CXX=g++3.3
CPP=cpp3.3 before compiling. The one reason that could fail I see right
now is if the gcc versions use different archspecs.

-- 
Arwed v. Merkatz                              Source Mage GNU/Linux developer
                                                    http://www.sourcemage.org



More information about the SM-Discuss mailing list