On Wed, Jun 15, 2011 at 15:39, Sam Ruby <[email protected]> wrote: >... > In the rare event that somebody wishes to take exception to a change > like this, we deal with that on an exception basis; sometimes this is > simply a matter of somebody else fixing the fix. In extreme cases we > may decide to first revert the controversial change and then talk > through the issue on the relevant list.
Short note: ALWAYS let the original person perform the revert. They should recognize lack of consensus, understand there isn't a way forward at the time, and perform the revert. (of course, the "wrong" stuff could be left in, the community decides a fix, and you move onwards without a (temporary) revert) The only reason to perform a revert for somebody else's work is if there is build breakage that prevents everybody from working (and note that I said *build* rather than *test*). Or maybe if the person just drops off communications for an extended period of time. Taking a unilateral action (revert) against somebody else's commit is one of the highest forms of antisocial behavior. I've run into this a couple times[1] and so it really wanted to stress this particular point. And yes, I know Sam wasn't referring to this kind of behavior. He said "we may decide". I just wanted to clarify who would *perform* the revert after that decision. Cheers, -g [1] the latest was actually a commit/revert war that went two cycles before it escalated. very uncool.
