Hi Mathieu, Apologies if gmail mess up email quoting.
On Thu, 5 Mar 2020 at 10:33, Mathieu Lirzin <mathieu.lir...@nereide.fr> wrote: > > PS: Votes in regards code modification in a community with PMC members > that basically disagree on everything else than “We are an awesome > community” is not a solution to anything because any veto with a valid > technical decision like for example “JMockit is redundant with Mockito > purpose which is already used” would simply block your proposition, so a > vote for code modification serves only to validate a consensus. [1] > > [1] > https://www.apache.org/foundation/voting.html#votes-on-code-modification > > The link regarding voting is helpful, thank you. It gives me an insight into how decisions get accepted. We already have a +1 from James. Can I count on your vote, Mathieu? :) Regarding the example argument “JMockit is redundant with Mockito purpose which is already used”: I hope I've demonstrated in this thread that Mockito is not fit for purpose in this particular case due to the architecture of the MacroFormRenderer class; and that refactoring MacroFormRenderer without the support of tests is risky and therefore a worse proposition that introducing JMockit. Therefore a veto with that particular argument shouldn't be considered valid. Also, since we are only discussing test code, we can easily remove them later if sentiment goes against JMockit. > > with a view to retire and replace them with more standard and > > recognisable Mockito tests once made possible through refactoring. > > I would suggest seeking the goal of adopting basic dependency injection > with explicit method/constructor arguments which leads to simpler tests > and is a sign of well designed code, instead of seeking the replacement > of advanced mocking techniques with less advanced ones. Mocking is > sometimes useful but should be avoided when possible. > I'd happily support this and leave the JMockit tests in place if that is considered reasonable after refactoring the class to introduce dependency injection. But I also wouldn't stand in the way of later conversion to Mockito if the community wanted it. In summary, I'm asking the community to accept introduction of a new mocking library to support a longer term piece of refactoring work. Committers! Bring me your +1 and lets get these PRs merged :) Thanks, Dan. -- Daniel Watford