bzr-svn tests on Windows

[Martin Pool]

2009/6/16 Maritza Mendez <martitzam at gmail.com>:
>
> Testing is obviously important.  And I think we've seen that several recent
> releases of bzr (not to mention plugins) might have benefited from a more
> coordinated approach.  But I respectfully suggest that the core team has
> shown commendable openness and transparency.  So rather than speculate on
> what Canonical is or is not doing on its own, perhaps some of the rest of us
> need to step up more.
>
> For my part, I am willing to do end-user testing of anything and everything
> about "bzr on Windows" subject to only three simple and reasonable
> understandings:
>
> 1.  I set my own schedule, as this is a labor for passion not pay.
> 2.  The developers continue, as they already do, to patiently explain how to
> install the software, including dependencies.
> 3.  The test methods and any special scripts or tools are provided by the
> developers.
>
> I see a lot of progress recently, especially in the areas of release
> management and issue triage.  Few FOSS projects can claim the commitment to
> quality and transparency that the bzr core developers show every day.  Many
> good things are happening and more are needed.  That takes a user community
> willing to step up and fill gaps.

Thanks very much Maritza, both for your kind mail and testing.

Servers for this stuff are nice, but just setting up machines doesn't
in itself fix anything: interactive testing is needed as well as
scripts, especially for release candidates, and just running the tests
doesn't fix the failures.  As we discussed in Maritza's other threads,
it can be hard to close the loop between a user getting a bunch of
test failures and a developer making incremental progress, but we will
get there.

Vincent is working on a buildbot setup as a background task, and I'm
working on getting a VM test facility supported by IS, which is a bit
more complex than buying consumer hardware.

-- 
Martin <http://launchpad.net/~mbp/>