2011/4/14 Mark A. Hershberger <mhershber...@wikimedia.org>: > I think branching 1.18 immediately after the 1.17 release (or now, for > that matter) will help us manage code review better. If we have people > testing the 1.18 branch and updating regularly (similar to what Ubuntu > does for their development) and we set a date (July 15th?) when we know > we have to have a release prepared, then that will help Code Review all > the more. > July?!?
I know 1.17 took a long time, but that was like a year's worth of code. We should strike to keep the branch-to-release time as low as we can, and it definitely needs to be WAY less than 3 months. It's been like 4 months for 1.17, but 1.17 was quite exceptional, and more frequent and quicker releases should become the rule. My opinion is it would be best to branch 1.18 now-ish and revert Happy-melon's Action changes (he wholeheartedly agreed that's 1.19 material). Slightly off-topic: Also, we should get our code review act together in a more sustainable way. I've brought this up before, but it hasn't gotten a lot of attention, probably due to the 1.17 craze. We have to have a serious discussion about code review reform (to use a political-sounding term); I think the tech staff meeting after the Berlin hackathon would be a good venue for discussing the WMF side of this. The conference itself is really supposed to be a hackathon this time, so I'm not sure that having a protracted discussion there would be a very good idea; that's basically what we did the whole time last year, and this year is supposed to not be like that for a reason. As always we do of course need to be careful to not want to solve this "internally" between WMF staff, but have a public discussion with everyone regardless of whether they happen to be paid. However, my impression is that this particular topic is one that mainly involves staff and that it would be acceptable to hammer something out internally and propose that on wikitech-l as something of a draft, in this particular case. I'd be very interested to hear how unpaid developers feel about that, as some of them have called out this practice as undesirable back in September. Roan Kattouw (Catrope) _______________________________________________ Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
