On Thu, 2010-07-15 at 13:03 -0400, Will Woods wrote:
On Thu, 2010-07-15 at 07:17 -0400, Kamil Paral wrote:
> == Proper solutions ==
> 3. Pass all autoqa variables into the control file as a dictionary,
> instead of a pile of variables. We have talked about this some time
> ago. It will need a modification of all the test cases and some
> tweaks in autoqa harness.
Right. We might be able to cheat a little by changing the control files
to pass the dict as keyword args, like so:
-job.run_test('classname', name=name, config=autoqa_config, ...)
+job.run_test('classname', **autoqa_args)
If that works as expected, we can avoid having to change the test
wrappers... but see below.
> But overall it shouldn't be such a hard change. Vojtěch could do it
> easily I believe.
>
> What do you think?
Sounds like a good idea, and I think now's a good time, since we're
already making a whole lot of little changes to the tests for Josef's
AutoQATest baseclass patch.
These two patches will change the test code/examples fairly
significantly, so I'm thinking we should bump the version number to
0.4.0 to reflect that. Which means we might want an 0.3 branch in git,
so we can keep the current production code bugfixed / up to date while
we make these changes.
Does that sound reasonable? James, you're the current buildmaster - is
it useful/necessary to keep a branch for 0.3.x? Do you anticipate us
needing 0.3.x bugfixes in the next week or two while the 0.4.0 code
shapes up?
The only 0.3.0 bugfix on my radar is the issue Kamil discovered with the
koji-build watcher and repoinfo. I'm still investigating a fix for
that, but I'd be just as fine landing that in 0.4.0.
If we had to create a branch for some reason, it's not too much trouble
to create a branch from a specific point in the master branch history,
right? Something like ...
# git checkout -b v0.3 a200e6a49d47a5b9242f11858315f26ad0a3381e
Thanks,
James