[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: svn branches/1.4.x r25941: FAIL (i686-debian-sarge1 shared gcc-3.3.5)

From: Erik Huelsmann <ehuels_at_gmail.com>
Date: 2007-08-03 10:43:58 CEST

On 8/3/07, Lieven Govaerts <svnlgo@mobsol.be> wrote:
> Quoting Erik Huelsmann <ehuels@gmail.com>:
>
> > On 8/3/07, buildbot@mobsol.be <buildbot@mobsol.be> wrote:
> > > Full details are available at:
> > >
> >
> http://www.mobsol.be/buildbot/i686-debian-sarge1%2520shared%2520gcc-3.3.5/builds/1862
> > >
> > > Author list: djh,dlr,dws,glasser,hwright,lgo,vgeorgescu
> > >
> > > Build Slave: eh-debsarge1
> >
> > Lieven,
> >
> > The build bot removes the 'build' directory when switching to a
> > branch. This means that python can't set the CWD for the cleanup
> > process to ..../build. Shouldn't we just call the cleanup script from
> > .../ instead of from build, giving the path to be cleaned as an
> > argument? That way, we hopefully can avoid these bogus failures.
>
> The alternative I was thinking of is to set flunkOnFailure to False on the
> cleanup step. While the cleanup step can still fail, the resulting build will
> still pass. Check the buildbot master config:
> http://www.mobsol.be/svn/public/buildbot/trunk/config/masterdir/master.cfg. I
> currently only set flunkOnFailure to True for update, build and test, but that
> seems to be the default.

I actually think that's a tough one: if cleanup fails, you can get a
successful build just because of it...

How about adding a workdir="" parameter? Then we can pass a "build" as
a parameter to svnclean.sh to signal that it should clean the "build"
dir, if available.

bye,

Erik.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Aug 3 10:42:22 2007

This is an archived mail posted to the Subversion Dev mailing list.