On Mon, Sep 16, 2013 at 1:05 PM, Daan Hoogland <daan.hoogl...@gmail.com> wrote: > A worthy cause, and we should all strive for it. It does put extra > constraints on things like db upgrade scripts however. If ACS is > releasable at all times, can users make their own release and expect > all upgrade paths to work? > > I disagreed with you on documentation as a commit criteria in another > thread, David. I do favor (extensive) unit tests as criteria.
Disagreeing is fine - no one (least of all me) says that my opinion is right all of the time. :) Over time I've become less concerned about a feature working, and more concerned about changes in a feature breaking the rest of CloudStack. Both are important - but I am to the point that I think that new features (and bugfixes) need to demonstrate that they don't harm the rest of ACS in the process. The real problem is that CloudStack is so large in scope that not many understand the full scope of interactions. --David