openstack team mailing list archive

[Monty Taylor <mordred@xxxxxxxxxxxx>]

On 01/03/2012 02:11 PM, Jason Kölker wrote:
> On Tue, 2012-01-03 at 21:38 +0000, Mark McLoughlin wrote:
>>> As a related note, I'm going to get the current repo moved in to gerrit
>>> today or tomorrow.
>>
>> It's more Jason's call, but I think we're basically asking you to hold
>> off on that for a little while. We may decide to start a new repo.
> 
> I've got Melange going off a non-master branch from my github repo at
> the moment, so we could start with a clean repo and not have to worry
> about breaking melange while we figure out the api on our side should
> look like after surveying the other projects.
> 
> I'm 50/50 on the blank slate approach. I'm leaning towards just taking
> what is there as is, and just start refactoring it as needed. Since
> melange is running off the other branch, we won't have to worry about
> compat at first and can make sweeping changes.
> 
> One thing I would like to ask is if we could not use
> run_tests.sh/run_tests.py in the jenkins job as I'd like to remove it.
> Currently in the setup.cfg there is a note on how to get the same output
> (style wise) with nosetests. The only downside is I haven't written a
> nose venv plugin yet, but if we move it to tox, then that shouldn't be
> needed anyway.

I'd be for that 1000% and will work with you on making it happen not
just in openstack-common, but across the board. I have been wanting to
kill run_tests.py since the first day that we imported nova into the
code repos. If we can get this tox off the ground, then we're
potentially in GREAT shape.

BUT - you already have the nose plugin and the pep8 checker done, so
consider me sold on that! :)

Feature request - can you add an option to tissue that will cause it to
output in pylint-compat format? (because we already have a jenkins thing
that reads that output. In our jenkins jobs, we do this:

pep8 blah | perl -ple 's/: ([WE]\d+)/: [$1]/' > pep8.txt

Which is a silly filter, but I'd love to have less stupid things in
jenkins jobs.

Also, have you seen the way that py.test displays tracebacks?

Monty