#general
- b
bright-apartment-76721
02/07/2017, 8:25 PMso I guess in pure outcome space, in my ever-recurring example of the 100k scala files that don't really depend on one another in awkward ways, all stuffed into a single target, with a shared cache - b
bright-apartment-76721
02/07/2017, 8:25 PMwould the majority of the 100k files get compiled more than once across a team using the shared cache? - b
bright-apartment-76721
02/07/2017, 8:25 PM(assuming they mostly don't change) - w
witty-crayon-22786
02/07/2017, 8:25 PMwho is writing to the shared cache - w
witty-crayon-22786
02/07/2017, 8:26 PMand are they using
, as i mentioned above--no-compile-zinc-incremental - b
bright-apartment-76721
02/07/2017, 8:26 PMlet's just say the poor soul who had to build all 100k files the first time is the only one writing to the cache - b
bright-apartment-76721
02/07/2017, 8:26 PMand sure, I can disable zinc if that helps in this scenario - b
bright-apartment-76721
02/07/2017, 8:26 PMas I said, this isn't actually happening in practice 🙂 - w
witty-crayon-22786
02/07/2017, 8:27 PMif they do a build immediately after
or they passclean-all
, then that build will be written to the cache--no-compile-zinc-incremental - w
witty-crayon-22786
02/07/2017, 8:28 PM(again, as i mentioned: this is why we disable incrementalism in CI: to ensure that everything that CI builds is written to the cache) - w
witty-crayon-22786
02/07/2017, 8:28 PMif you disabled incrementalism locally, then zinc would... not be incremental. but everything would be written to the cache. - b
bright-apartment-76721
02/07/2017, 8:29 PMyeah, that makes sense; I'm more asking about these things reading from the cache to avoid work though - w
witty-crayon-22786
02/07/2017, 8:29 PMyou read from the cache if the fingerprint for the files matches an existing artifact in the cache. - b
bright-apartment-76721
02/07/2017, 8:30 PMaha! I think that's what I was missing. So in this case, since most of the files are unchanged, I would read from the cache and get the associated pre-built .class file? - w
witty-crayon-22786
02/07/2017, 8:30 PMthe cache is on a per-target basis. - b
bright-apartment-76721
02/07/2017, 8:30 PMoh okay - w
witty-crayon-22786
02/07/2017, 8:30 PMbecause that's the level that zinc works at. - b
bright-apartment-76721
02/07/2017, 8:30 PMso I wouldn't unless I built the exact same thing on a different machine - w
witty-crayon-22786
02/07/2017, 8:30 PMcorrect. - b
bright-apartment-76721
02/07/2017, 8:31 PMcool, I think I finally get - b
bright-apartment-76721
02/07/2017, 8:31 PMit - b
bright-apartment-76721
02/07/2017, 8:31 PMthanks, and sorry for being dense! - w
witty-crayon-22786
02/07/2017, 8:33 PMand to be clear: the "was it incremental?" decision is thus also on a target-by-target basis - w
witty-crayon-22786
02/07/2017, 8:34 PMso editing some root target (a
target, for example) only affects the cache key of that target, and only influences whether the artifact for that editted target is writtenjunit_tests - b
bright-apartment-76721
02/07/2017, 8:48 PMI see - r
rough-dog-96241
02/07/2017, 10:04 PMIs groovy a thing in pants? Asking for a friend. 🙂 - w
witty-crayon-22786
02/07/2017, 10:07 PMit's not - r
rough-dog-96241
02/07/2017, 10:14 PM👍 - w
witty-crayon-22786
02/07/2017, 10:17 PMadding groovy support that depended on java/scala would not be insanely difficult. but going the other way (java depending on groovy) would be challenging, because
"owns" thezinc
product, and bouncing back and forth between groovyc and zinc would require an overhaul.runtime_classpath - w
witty-crayon-22786
02/07/2017, 10:17 PMwe've been working on said overhaul for a while, but it's still a few months away.