Hi, On Thu, Mar 19, 2015 at 9:35 PM, Matthieu Moy <Matthieu.Moy@xxxxxxxxxxxxxxx> wrote: > Eric Sunshine <sunshine@xxxxxxxxxxxxxx> writes: > >> I also tend to favor adding "failure" tests which are flipped to >> "success" when appropriate, however, as this is an entirely new >> feature, this approach may be unsuitable (and perhaps overkill). > > I can buy the "overkill", but not unsuitable. Even for new features, the > tests should fail before and pass after. Otherwise, the tests are not > testing the feature. Actually, this is a strong principle in test-driven > development: don't write code unless you have a failing test. > > But I was just thinking out loudly, certainly not requesting a change. I also think that the tests belong in their own patch as the patch is really long compared to the rest of the patch series and it makes it easier for me to respond to comments as well. Putting the tests before the implementation, though, makes sense in this case as it is just an implementation of a well-defined set of requirements. Some tests will still pass with or without the feature, but that is the requirement of the feature -- which is to not be enabled until the user explicitly creates the xdg file. Will re-order the patch series in the next version. Thanks. Regards, Paul -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html