On 08/03/14 02:48, Bartosz Dziewoński wrote:
On Fri, 07 Mar 2014 16:27:53 +0100, Antoine Musso hashar+wmf@free.fr wrote:
So a single -1 should prevent a change from being submitted until that -1 is lifted by addressing the person concern(s) or correcting him/her or whatever.
Note that such a rule never been followed by anyone, including by various WMF teams. You can easily find numerous examples, even if these searches are limited to only changesets where the -1 stuck on the last patchset.
https://gerrit.wikimedia.org/r/#/q/is:merged+label:Code-Review-1,n,z (this search seems to hang forever) https://gerrit.wikimedia.org/r/#/q/is:merged+label:Code-Review-1+project:med...
I wouldn't go so far as to say nothing with a -1 should ever be merged. I think objections should be resolved when the objecting party is actively involved and competent, and when the objection is substantial.
That is to say, if someone drops a -1 on a change, but despite their objections being answered in a subsequent comment or patchset, they have no further involvement, then the change can be merged after some time has elapsed.
And if someone complains about some minor creative element, such as whitespace, function naming, comment punctuation, etc., then that can be resolved without requiring consensus, e.g. by simple majority or by respecting the decision of the original author.
Or if someone is commenting outside their field of expertise, and their objections are wrong by widely-accepted matter of fact, those objections can be ignored.
But if someone is saying that the entire change is a bad idea, and has some solid reason for saying that, and has renewed their objection as appropriate, then I think that needs to be resolved by further discussion.
-- Tim Starling