Le 01/01/2018 à 13:13, Michael Brohl a écrit :
I won't insist on this further. There seems to be a substantial difference in perception how we should handle these issues and I'm tired to
discuss this again and again. Hope you will take care about the fixes resulting from the commit if they are necessary.
Yes as I said I'll review again. I hope the contributors (Suraj and Anushi )
and others will help me do so, it's indeed a large task
I would expect that the testing was done before providing the patches. You cannot rely on the tests in all cases because different return results
might not affect the single event/service but the integration logic/flow. I am not sure if we have enough tests for this.
The integration tests are/have-been mostly designed with the functional aspect in mind. I don't think we have much tests by service, if any. And yes
we have not enough tests at large, but we have already 1307 integration tests and 26 Gradle unit tests
Despite them it's really hard to test only. We have almost to decide on each change. For instance should we keep an continue in a previous way or not.
Sometimes it's not obvious. See my last (today) comments in OFBIZ-9981 for instance
Jacques
