On 24/06/13 at 10:15 +0300, Arto Jantunen wrote:
> Lucas Nussbaum <[email protected]> writes:
> > On 23/06/13 at 22:19 +0300, Arto Jantunen wrote:
> >> Lucas Nussbaum <[email protected]> writes:
> >> > On 23/06/13 at 21:19 +0300, Arto Jantunen wrote:
> >> >> I can't reproduce this, both sbuild and pbuilder build this successfully
> >> >> on current unstable. It seems that either there was a problem with one
> >> >> of the build dependencies that has been fixed (none of the obvious ones
> >> >> were different between this build and mine, as far as I can see) or
> >> >> there was something wrong with the build host (running out of disk space
> >> >> comes to mind first). Could you retry this?
> >> >> 
> >> >> doc/exts/xmlschema.py is apparently known to have crappy error handling,
> >> >> so getting a useless exception seems to be expected when this fails. If
> >> >> the failure is reproducible it should probably be patched a bit to
> >> >> figure out where the actual problem might be..
> >> >
> >> > Hi,
> >> >
> >> > Could you provide your build log (preferably with sbuild) and a diff
> >> > between your log and my log?
> >> 
> >> Log from my latest sbuild run (I did about a dozen to check for a random
> >> failure, all successful) is attached. I did run diff on these, but
> >> couldn't find anything obvious (there are a lot of differences).
> >> 
> >> Hmm. Does the vm happen to have network access disabled? That could be a
> >> possible source for the difference. Sphinx has a tendency to download
> >> stuff from random network locations, and my build machine has normal
> >> network capabilities (and no local mirror).
> >
> > Erm. What kind of stuff does it download?
> 
> If the catalogs it uses to generate crossreferences aren't available on
> the system it will download them. This is the source of most of Bcfg2's
> build-deps. As far as I'm aware this can't be disabled without ugly
> kludges.
> 
> > I retried, and could reproduce the failure. The machine has normal
> > internet connectivity.
> 
> This is interesting indeed. Did you notice anything useful in the diff
> between your build log and mine? If nothing turns up there I should
> probably look at fixing the error handling in xmlschema.py..

no, I didn't find anything...

Lucas


-- 
To UNSUBSCRIBE, email to [email protected]
with a subject of "unsubscribe". Trouble? Contact [email protected]

Reply via email to