Open Stack

On Wed, 27 Feb 2013 06:57:22 -0500 (EST)
Attila Fazekas <afazekas at redhat.com> wrote:
> I agree we must stop skipping tests based on weather conditions or
> something like that. (But, We can keep and/or extend the early
> failure logic.)
> 
> The other option would be using attributes or coding patterns to the
> testcase's name. (nosetests -x ec2 # skips the ec2 tests for example)
> We could skip tests based on path as well.

Personally I'd prefer that to be specified in the tempest config file
because I think it would be easier for people new to tempest or who run
tempest only occasionally to understand and setup.

> Any not skip exception in setUpClass is considered as an ERROR or
> FAIL (if the class has non zero test_ method).
> 

Yes you're right with nosetests - I thought I had checked, but
obviously got it wrong. With testr it does definitely appear to still
PASS though. I'm not sure if I'm just doing something wrong or not, but
I guess setupClass is going away in the long term anyway.

> We should do the skipping based on service endpoint expected to work
> or not, if we doing it on a configuration level.
> 

Yes, I agree.

Regards,

Chris
-- 
cyeoh at au1.ibm.com