Ergonomic, hermetic, user-friendly software build system for Python, Go, Java, Kotlin, Protobuf, Scala, Shell. Pants lets you fearlessly scale up your codebase!

Pants

it’s not. I’m building bazel to play around with kythe.

I’m not sure I ever fully understood the value prop of kythe et al

If you have a few hours, you should discuss that with benjy ;)

<@U053HNE90>: any insight into these jvm-related CI issues? seem to recall a recent discussion around jvms in CI but not much more context than that

```-------------- Captured stderr call --------------
                     error: error reading /home/travis/.ivy2/pants/junit/junit-dep/javadocs/junit-dep-4.10-javadoc.jar; error in opening zip file```

I'm not sure what, if any, concurrent sharing goes on in Travis

going back to old commits that passed and re-running their shards seems to result in the same failure which feels odd unless its rebasing against current master - but would be totally unexpected

wonder if travis made some nonvisible change on their backend

Seems like it must be a caching issue.  Or in some bizzarro universe, maven central is serving up a corrupted zip

welp, clearing the master cache seems to have fixed at least one instance of it

rerunning the failed shards for the last two failed master commits

i read about travis at some point, they are using a docker app for each shard

so the cache should be self contained, inaccessible from other shards

&gt; working bazel built on OSX when we did the bakeoff.
Was that an internal bakeoff or is there a cool google doc floating around somewhere?

My experience with Bazel so far has been incredibly googly (and not in the good sense)

Although admittedly a lot of that was attempting to debug the kythe build, but bazel certainly doesn’t make that easy

<@U0AALHEDR>: the cache is stored somewhere after each run finishes, and restored before the next one starts. so things do live between runs (otherwise it wouldn't be much of a cache)

... but i guess you mean that different shards within one run shouldn't be able to see the other's caches... and that should be true

four twitter shipits ... anyone else able to eyeball <https://rbcommons.com/s/twitter/r/3445/> ?

I assume you mean somoen not from twitter :simple_smile:

there is a general consensus among team leads (but not a documented rule) that shipits should span organizations