#general
- m
mammoth-autumn-2551
08/03/2015, 2:35 AMFWIW this behaves as I would expect in the pants repo with the
from pants HEAD, (the target is invalidated correctly if I edit a source file)./pants run ./examples/src/python/example/hello/main - m
mammoth-autumn-2551
08/03/2015, 2:35 AMChanges to sources - m
mammoth-autumn-2551
08/03/2015, 2:36 AMBut it does reproduce in our repo in a very indirect way, so I'm gonna drill in and try to get a smaller repro going - a
astonishing-jelly-60479
08/03/2015, 2:22 PMis there any documentation on artifact caching or the right way to set up pants in a multi-worker CI environment? we'd like our workers to share as much as possible and i'm wondering what's safe to share, stuff like the --pants-workdir/--pants-supportdir - b
bored-art-40741
08/03/2015, 2:22 PMNeither of those are safe to share - a
astonishing-jelly-60479
08/03/2015, 2:22 PMactually share is the wrong word, sorry - a
astonishing-jelly-60479
08/03/2015, 2:23 PMre-use between runs - b
bored-art-40741
08/03/2015, 2:23 PMThere are a couple of google docs floating around on how to setup regular artifact caching - b
bored-art-40741
08/03/2015, 2:23 PMAh - b
bored-art-40741
08/03/2015, 2:23 PMSo .pants.d is safe to reuse, for some definition of safe that is largely task dependent - b
bored-art-40741
08/03/2015, 2:24 PMArtifact caching is more robust and designed specifically to be safe to share and location agnostic, but once again some tasks have had more battle testing than others - a
astonishing-jelly-60479
08/03/2015, 2:24 PMnoted. thanks. - b
bored-art-40741
08/03/2015, 2:26 PMFor us, the most valuable caching is the results of JVM compilation, particularly scala - b
bored-art-40741
08/03/2015, 2:26 PMWe also get some value out of caching codegen results and buildgen intermediate analysis - a
ancient-stone-50795
08/03/2015, 2:26 PM@astonishing-jelly-60479: I just shared a doc with you - a
astonishing-jelly-60479
08/03/2015, 2:28 PMok, neat, thanks - a
astonishing-jelly-60479
08/03/2015, 2:28 PMnot sure how much of a benefit we'd see in our mostly-python environment but worth playing with, maybe - b
bored-art-40741
08/03/2015, 2:28 PMWe should definitely codify these floating docs into proper documentation with a couple of case studies - b
bored-art-40741
08/03/2015, 2:29 PMWhich tasks do you find slow in your usages, @astonishing-jelly-60479 ? - a
astonishing-jelly-60479
08/03/2015, 2:30 PMhard to say right now, in the middle of a major upgrade and just trying to get as much configured correctly out of the gate - a
astonishing-jelly-60479
08/03/2015, 2:30 PMand ensure any old hacks are either unnecessary or properly replicated - b
bored-art-40741
08/03/2015, 2:30 PMAh okay. Yeah I wouldn't prioritize artifact caching early on in a mostly python environment - a
astonishing-jelly-60479
08/03/2015, 2:32 PMyeah, i'm not - a
astonishing-jelly-60479
08/03/2015, 2:32 PMalthough i can see some use for it if i can wrap our static asset build in it to save time on web builds that don't touch JS/CSS - b
bored-art-40741
08/03/2015, 2:34 PMAh yeah - b
bored-art-40741
08/03/2015, 2:34 PMNailing down the various ways different shops do web assets is something that came up at the summit - b
bored-art-40741
08/03/2015, 2:34 PMWe don't have a good public facing story on how to do it in a pants-ey way - a
astonishing-jelly-60479
08/03/2015, 2:36 PMour current method is not great, we have a custom target that tries to determine if it's being instantiated in the binary goal, builds assets at that time and adds them to the target resources - a
astonishing-jelly-60479
08/03/2015, 2:36 PMi'd like to make it more formal, a new target/task and use resource targets or something - b
bored-art-40741
08/03/2015, 2:40 PMYup, we are similarly hacky in our approach, and we'd also like to move over to generating synthetic resources