On 10/30/2013 07:56 PM, Jan Rusnacko
wrote:
Hi Jan,Hello Thierry, layout OK. As for tests - instead of reinventing the wheel by defing class Test_standAlone to set up instance, use py.test fixture. Also, you should not force setup, test, teardown execution for each test by specifying sub-methods for each test. Testing framework (py.test) should be doing that. I think this will make your tests fail badly if some exception occurs - if _test_ticket47560_setup raises exception, it will propagate back to py.test and cleanup method will never be executed for that ticket. thanks for you comments. py.test will call for for each test_ticketxxx setup setup will create the instance if it does not already exist, else it provide a dirsrv to it. teardown will remove the instance if the test_ticketxx is not able to properly clean it up (if test_ticketxx raise the clean_please flag) test_ticketxxx will start it executions with clean_please=true (set by setup), if it succeeds it set clean_please=false. Unless it fails to properly clean up the instance. In that case it sets clean_please to true. What would be the advantages to make the setup/teardown sub-methods of the test_ticketxx ? At least a "small" drawback is that the person that write test_ticketxxx will have to write them, instead of using the "standard" one. Also, I believe each ticket should have its own file which contains one or more testcases. I think that would reasonably group relevant things together. Right, it was my first intention. But then I had this idea to group the tickets per deployment module. I don't know if it is a good idea, but it seems to be confusing :-[ On 10/30/2013 05:57 PM, thierry bordaz wrote:https://fedorahosted.org/389/attachment/ticket/47578/0001-Ticket-47578-CI-tests-removal-of-sudo-and-absolute-p.patch -- 389-devel mailing list 389-devel@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/389-devel |
-- 389-devel mailing list 389-devel@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/389-devel