Fabian Groffen wrote:
> On 03-06-2008 09:39:46 +0200, Fabian Groffen wrote:
>   
>> Since we're rsync based, I can avoid taking an entirely new snapshot
>> (with the risk of it being broken) but just update certain things only.
>> So I guess I'll do that in this case to add in OpenSolaris support.
>>     
>
> I already added a new snapshot yesterday, but I found an annoying
> problem in gcc-4.2.3 that is fixed in 4.2.4 that is exposed during the
> bootstrap process.  Todays snapshot should finally be it!
>
> (This is mutt from my OpenSolaris/Solaris 11 machine with Vim)
>   
Nice - haven't personally tested the new snapshot, but I was able to 
bootstrap on OpenSolaris 2008.05 the other day using my own snapshot. 
Just for an experiment, I unmasked GCC 4.3.1 at the very beginning and 
was able to bootstrap and emerge the entire system using GCC 4.3.1 (as 
well as GCC 4.2.4)

My OpenSolaris install was very "clean" and I only installed SUNWgcc and 
SUNWgmake (+ dependencies) from the pkg.opensolaris.org repository.

If anyone else is interested, here is a couple of deviations from the 
Solaris 10 bootstrap doc I had to perform:
- "emerge --oneshot --nodeps =sys-devel/m4-1.4.10-r3" before emerging 
flex in CL 1.8 (perhaps the upstream flex ebuild should depend on m4 - 
or would they not bother because m4 is assumed to be a system package?)
- "emerge --oneshot --nodeps bison" before emerging GCC
-  when testing a bootstrap again with just GCC 4.3.1 I also emerged gmp 
and mpfr before gcc

Matt.


-- 
gentoo-alt@g.o mailing list