Code review comment for lp:~bilalakhtar/unity/5.0series-sru-software-center-integration-fixes

Hi Didier, it's not a surprise that the manual tests were not enough because they were very sparse before (a single test and it tested only a cancelled installation). In other words, it didn't cover anything of the functionality that had regressed. Now, with my updates, they specify multiple test cases that include every detail that should be verified.

So I'm not sure the tests as updated can still be considered "clearly not enough", unless of course for some reason these aren't run or used, in which case I'd agree that no manual test will be useful.

The main benefit of manual tests for this feature is that the feature spans two different applications. We have full automatic tests on the Software Center side, but that is not sufficient to also test the Unity side, and the same will be true with automatic tests on the Unity side. The best we can test is the API between then, but whether the two operate correctly can be very effectively tested by a person.

I'm not saying that automatic tests should not also be included (automatic unit tests are always veryimportant), but assuming these manual tests are actually used and run, the set that I've added will be a very significant enhancement to any unit tests and should be sufficient to detect any regression on either side for this shared feature.

Thanks!