Re: Bug#247211: [patch] Building python on hurd-i386
(putting debian-hurd back on the CC list, which I somehow messed up on
submitting the bug)
On Wed, May 05, 2004 at 03:30:54PM +0200, Marcus Brinkmann wrote:
> At Mon, 3 May 2004 22:24:24 +0200,
> Michael Banck wrote:
> > I finally took the other way and removed cthreads from the build
> > process. That seems to work fine now, python and python-tk build and
> > link without problems and seem to be functional (I installed reportbug
> > and it works fine except that I've got no MTA configured. I don't have X
> > running, so I can't test python-tk beyond trying to import it I guess)
>
> python probably wants to keep cthread support for native mach targets,
> I wouldn't know. So this could be a problem.
Ah, I did not think about this. In that case, there should probably be a
configure option like --disable-cthreads, to avoid mangling them with
pthreads.
> Apart from this, it is probably ok to require pthread on the Hurd,
> although I still consider the pthread support in the Hurd to be very
> quite experimental.
So perhaps we should keep this in the Debian package for now and see how
people get along with python2.3/pthreads? I can only say it does not
seem worse than cthreads. I've ran the test-suite manually yesterday,
and there was only one spurious segfault left, as opposed to three with
cthreads. I will try to post a further analysis soon.
If things works well enough, we can submit it upstream.
Michael
--
Michael Banck
Debian Developer
mbanck@debian.org
http://www.advogato.org/person/mbanck/diary.html
Reply to: