Doesn't close the "also good would be a supervisord event listener" bit of #197
N.B.: Changes to the layout of habitat.yml will require care when deploying.
test_startup.py unit tests are of questionable quality at best.
I believe that testing by stubbing out and mocking logger.Handler, logging.Logger, e.t.c., would be pointless (and is the reason so far startup hasn't had any tests) since it would just be a re-write of startup.py, but speaking mox. Instead of actual bugs we would then just have 'bugs' where the mocks we write don't behave the same as the real thing. Therefore, these tests mock input/output streams/emails and check written files to ensure that it sets up habitat logging correctly. I think (?) this is OK, though you might want to check it a bit more closely.
Doesn't close the "also good would be a supervisord event listener" bit of #197
N.B.: Changes to the layout of habitat.yml will require care when deploying.
test_startup.py unit tests are of questionable quality at best.
I believe that testing by stubbing out and mocking logger.Handler, logging.Logger, e.t.c., would be pointless (and is the reason so far startup hasn't had any tests) since it would just be a re-write of startup.py, but speaking mox. Instead of actual bugs we would then just have 'bugs' where the mocks we write don't behave the same as the real thing. Therefore, these tests mock input/output streams/emails and check written files to ensure that it sets up habitat logging correctly. I think (?) this is OK, though you might want to check it a bit more closely.