#general
- w
witty-crayon-22786
03/28/2016, 7:24 PMsure. the reason i differentiate between minor and patch is that i don't think this will be backwards incompatible - w
witty-crayon-22786
03/28/2016, 7:25 PMie, it's not necessary to use a minor. except for the release which makes the option the default... that's minor worthy - w
witty-crayon-22786
03/28/2016, 7:26 PMer... you're expecting that we'll only do validation for minor releases? i was figuring we do all the validation for patch releases as well - w
witty-crayon-22786
03/28/2016, 7:26 PMtbd, i suppose. - b
bored-art-40741
03/28/2016, 8:28 PMI expect minor to get significant QA from all parties before cutting the release. And to be honest, even for nominally backwards compatible changes, I expect unexpected breakage. It's the rule rather than the exception for this kind of integration work. - b
bored-art-40741
03/28/2016, 8:29 PMSo I don't want to release patch (which I expect to get somewhat less scrutiny than minor) for backend reworks even if they aren't visible changes. There's too much room for bugs. - b
bored-art-40741
03/28/2016, 8:29 PMBut yeah, TBD considering this is all from a mental model of our branch/release model that might not be what we go with - w
witty-crayon-22786
03/28/2016, 8:31 PMyea, that probably makes sense. - w
witty-crayon-22786
03/28/2016, 8:31 PMwould like to discuss it more though... obviously very relative on a change-by-change basis - w
witty-crayon-22786
03/28/2016, 8:31 PM(...whether something deserves minor/patch.) - b
bored-art-40741
03/28/2016, 8:31 PMyeah - w
witty-crayon-22786
03/29/2016, 12:05 AM@ancient-stone-50795, @jolly-chef-92794 : have you guys considered using
? should work fine for java codestrict_deps - a
ancient-stone-50795
03/29/2016, 12:27 AMEvery time I mention it to our site lead, I get a 10 minute lecture on why he thinks its a bad idea. On the other hand, he loves pants. - w
witty-crayon-22786
03/29/2016, 12:31 AMlol - w
witty-crayon-22786
03/29/2016, 12:31 AM@ancient-stone-50795: 2x compile time increase on some of our deeper targets - w
witty-crayon-22786
03/29/2016, 12:31 AMer, decrease. - a
ancient-stone-50795
03/29/2016, 12:35 AM@witty-crayon-22786: why is that? Is it some kind of inefficiency in zinc/sbt that gets faster? Or just a smaller classpath to pass to the compiler? - w
witty-crayon-22786
03/29/2016, 12:35 AMsmaller classpath - w
witty-crayon-22786
03/29/2016, 12:36 AMaffects both zinc and the compiler - h
happy-kitchen-89482
03/29/2016, 12:42 AMsmaller classpath == faster classloading, for one thing - h
happy-kitchen-89482
03/29/2016, 12:42 AMAnd zinc creates some internal data structures that are O(size of classpath) - h
happy-kitchen-89482
03/29/2016, 12:42 AMFor mapping external dependencies - w
witty-crayon-22786
03/29/2016, 12:45 AMwe'll likely do a one-time buildgen for java, and then enable it. - w
witty-crayon-22786
03/29/2016, 12:46 AMas a short term fix, we're also going to add "unused deps" errors/warnings - j
jolly-chef-92794
03/29/2016, 6:00 PMWe just hit a bug. https://github.com/pantsbuild/pants/blob/master/src/python/pants/backend/jvm/targets/credentials.py#L48 is using the wrong api for deprecated - j
jolly-chef-92794
03/29/2016, 6:00 PMIt’s passing a
as the first argument, which breaks becausebool
expects its first argument to be a callable predicatedeprecated() - w
witty-crayon-22786
03/29/2016, 6:01 PMblargh... good catch. - w
witty-crayon-22786
03/29/2016, 6:01 PM@jolly-chef-92794: are you all using netrc? - j
jolly-chef-92794
03/29/2016, 6:02 PMWell we reference it in a BUILD file. Not sure whether we actually need it still or not, @ancient-stone-50795 set that up. - a
ancient-stone-50795
03/29/2016, 6:03 PMYes, we do use .netrc to publish