#announce
- w
witty-crayon-22786
05/16/2016, 7:20 PMplaces to have multiple patch releases / rc cycles - w
witty-crayon-22786
05/16/2016, 7:22 PMand if there are only two at a time (
,1.0.x
for example) then you could call one conservative, and one aggressive1.2.x - w
witty-crayon-22786
05/16/2016, 7:22 PMand the aggressive one would essentially be fairly frequently retired - w
witty-crayon-22786
05/16/2016, 7:23 PM...so yea... thinking of it that way might be helpful. there is just already a lot in that page just to explainpre - b
bored-art-40741
05/16/2016, 7:24 PMStill not a fan of leaving a bunch of abandoned minor releases with no backport support - b
bored-art-40741
05/16/2016, 7:25 PMOkay how about this - b
bored-art-40741
05/16/2016, 7:26 PMconservative/aggressive strategy, where aggressive has RC cycles that cut off of master and become patch versions on the aggressive branch - b
bored-art-40741
05/16/2016, 7:26 PM3 month cadence for retiring aggressive to conservative, which is cherry-pick patched - b
bored-art-40741
05/16/2016, 7:27 PMIf you have a change that breaks backwards compatibility in such a way that a minor or major version bump is necessary, you must develop it on a branch until the next minor/major release cycle - w
witty-crayon-22786
05/16/2016, 7:27 PMehhhh. pretty not-down with long lived feature branches. - b
bored-art-40741
05/16/2016, 7:27 PMIn other words, even master is an inappropriate place to break backwards incompatibility if you can't do it in a way that has an escape hatch and defaults to staying compatible - b
bored-art-40741
05/16/2016, 7:28 PMIdeally there wouldn't be many or any - b
bored-art-40741
05/16/2016, 7:28 PMBut they would serve to discourage truly backwards incompatible changes - w
witty-crayon-22786
05/16/2016, 7:29 PMi feel like they are fairly well discouraged right now? - b
bored-art-40741
05/16/2016, 7:30 PMI mean, saying that you're forbidden from merging a change into master because it breaks backcompat is fine with me - w
witty-crayon-22786
05/16/2016, 7:30 PMbut also, i think the point of the conservative branch is to make that type of caution less necessary - b
bored-art-40741
05/16/2016, 7:30 PMBut it would be nice to have a system in place for moving those changes through if they're otherwise desirable - b
bored-art-40741
05/16/2016, 7:30 PMYes but the aggressive branch shouldn't break backcompat either - w
witty-crayon-22786
05/16/2016, 7:30 PM@bored-art-40741: this feels like a change of topic maybe - b
bored-art-40741
05/16/2016, 7:31 PMI think this is closer to the core of the issue - b
bored-art-40741
05/16/2016, 7:31 PMWe keep getting stuck on how various strategies represent backwards incompatibility wrt semver - w
witty-crayon-22786
05/16/2016, 7:31 PMi don't feel stuck on that. - w
witty-crayon-22786
05/16/2016, 7:32 PMthe primary issue with https://rbcommons.com/s/twitter/r/3890/ in that regard is that it burns minor versions unnecessarily - b
bored-art-40741
05/16/2016, 7:32 PMYes - w
witty-crayon-22786
05/16/2016, 7:32 PMbut it is still correct, in that it isn't necessary for a minor bump to break things. - b
bored-art-40741
05/16/2016, 7:32 PMI'd argue that we should be rate limiting our minor/major bumps - w
witty-crayon-22786
05/16/2016, 7:32 PMand if you did get consensus to cause breakage, the next stable branch created would clearly be a major bump - w
witty-crayon-22786
05/16/2016, 7:32 PMwhy? - b
bored-art-40741
05/16/2016, 7:33 PMBecause I don't like abandoning a bunch of minor versions with no backporting strategy in place - w
witty-crayon-22786
05/16/2016, 7:33 PM(bio, back in a bit.)