I submitted a series of patches the integrates python-nose with vdsm testing. I've been getting a lot of questions in the form of "Why?" so I decided to answer them.
The vdsm team has a bunch of tests written that help make sure that VDSM is functioning properly. Other Ovirt partners asked us to make the tests public and upstream and we were more then happy to oblige.
The tests internally used a custom made test infrastructure because the guy responsible for choosing the infrastructure said that there isn't a good open source alternative. This is of course not the case and I've been planning to move to a more standard testing framework so I wouldn't have to maintain a costume one. When Ayal asked if someone from our team could move all the tests upstream I volunteered to do it, mainly, because I thought it will be a good chance to finally change the test runner.
Our requirements are (In no particular order): * Integrates with python unit tests * Coverage support * Test skipping (Some tests are only valid for some configurations) * Configurable output * Integrates with Jenkins * Extensible using python * Already packaged for RHEL\EPEL and Fedora * Well supported and documented
nose answers all of these!
Also, an internal Red Hat survey shows that most teams here that work with python use nosetests and are very happy with it. Further more, Open Stack also use nose and have done a lot of tool integration. Seeing as we both use a similar toolset, we could share gerrit\Jenkins hooks.
Another thing I was also asked is why do the color wrapper around the output. When I first got my hands around the original test runner we had here I wrote coloring to it. For me, it's easier to skip a long list of tests and see the red pop out instead of looking for the word "FAIL".
vdsm-devel@lists.fedorahosted.org