https://pantsbuild.org/ logo
Powered by
#general
  • w

    witty-crayon-22786

    05/02/2017, 10:33 PM
    invalidates "all the things"
  • w

    witty-crayon-22786

    05/02/2017, 10:50 PM
    @gifted-autumn-88099 : agh, sorry for missing your question. it's possible to list multiple scopes, although it's not at all obvious: 
    scope='compile test'
    , for example.
  • g

    gifted-autumn-88099

    05/03/2017, 2:30 AM
    Awesome, thanks! I think that's exactly what I was looking for.
  • g

    gifted-autumn-88099

    05/03/2017, 2:30 AM
    Is a manual update to https://github.com/pantsbuild/pants/blob/master/examples/src/java/org/pantsbuild/example/README.md the right way to update the docs? Happy to send a PR.
  • e

    enough-analyst-54434

    05/03/2017, 1:19 PM
    @gifted-autumn-88099 - yup, that's the correct update procedure. Thanks a bunch and welcome back!
  • c

    clean-fish-15260

    05/03/2017, 5:55 PM
    how would I begin debugging why my OS X developer box isn't getting cache hits against our remote pants cache?
  • c

    clean-fish-15260

    05/03/2017, 5:55 PM
    like a very basic check has different hashes being checked from the exact same git repo checkout
  • c

    clean-fish-15260

    05/03/2017, 5:57 PM
    I don't think it's about maven resolution, unless it's about a corrupt / nondeterministic ivy cache on disk
  • c

    clean-fish-15260

    05/03/2017, 5:58 PM
    the remote cache is populated by Jenkins running on Linux, so I feel like that large difference is likely to be the core of the issue, but from reading here it seems like some of you all are on similar setups?
  • w

    witty-crayon-22786

    05/03/2017, 6:31 PM
    @clean-fish-15260: there isn't a fantastic way without edits to pants.
  • w

    witty-crayon-22786

    05/03/2017, 6:31 PM
    but looking at 
    -ldebug
    for cache keys, and seeing how low in your graph the disagreement starts might help
  • w

    witty-crayon-22786

    05/03/2017, 6:32 PM
    can look in local caches on CI and laptop hosts after a build of the same thing, and compare them
  • c

    clean-fish-15260

    05/03/2017, 6:32 PM
    I thought the JRE would have slightly different versions, but they actually match spot on
  • w

    witty-crayon-22786

    05/03/2017, 7:24 PM
    @clean-fish-15260 : we've never encountered a "linux vs osx" cache hit rate issue... nothing ends up in the fingerprint there
  • w

    witty-crayon-22786

    05/03/2017, 7:24 PM
    a few places to check: 1) options consumed by zinc being different
  • w

    witty-crayon-22786

    05/03/2017, 7:25 PM
    2) the contexts being built in CI and locally being drastically different (ie 
    ./pants test ::
    vs 
    ./pants test some/thing
  • w

    witty-crayon-22786

    05/03/2017, 7:25 PM
    various options are marked 
    fingerprint=True
    , so you can't change those in different environments.
  • w

    witty-crayon-22786

    05/03/2017, 7:26 PM
    ...and expect to hit
  • a

    adventurous-art-72916

    05/03/2017, 8:36 PM
    how to I get pants to use a different version of python?
  • a

    adventurous-art-72916

    05/03/2017, 8:38 PM
    I'm a little confused where it gets the version of python it uses. Pant's python doesn't support TLSv1.2, but my system python does
  • h

    high-advantage-47550

    05/03/2017, 8:39 PM
    raise TooManyCacheSpecsError('Too many artifact cache specs: ({0})'.format(spec))
    — hmm, is there any way/plugin to write to multiple cache locations?
  • r

    rough-sundown-61224

    05/03/2017, 8:41 PM
    @adventurous-art-72916 the python that pants itself runs as, for a plugin? It picks it at the top of the pants run script like 
    PYTHON=${PYTHON:-$(which python2.7)}
    , or the python it’ll use to run things in the repo, which is based on the interpreter_constraints setting.
  • a

    adventurous-art-72916

    05/03/2017, 8:43 PM
    the latter
  • r

    rough-sundown-61224

    05/03/2017, 8:43 PM
    @high-advantage-47550 the cache is pretty set on being either a local cache or a REST cache (potentially with a server set) backed by a local cache. You could write a cache plugin that does multiple locations, but you have to monkeypatch the logic for creating caches.
  • a

    adventurous-art-72916

    05/03/2017, 8:43 PM
    let me try interpreter constraints
  • h

    high-advantage-47550

    05/03/2017, 8:43 PM
    @rough-sundown-61224 makes sense. i’m actually trying to write to both a REST server and use your s3 plugin 😄
  • r

    rough-sundown-61224

    05/03/2017, 8:45 PM
    @adventurous-art-72916 my old laptop used to pickup pyenv which wasn’t friendly for a number of reasons, which might be what yours is doing.
  • r

    rough-sundown-61224

    05/03/2017, 8:46 PM
    It’s actually @adventurous-art-72916’s s3cache, creating a strange loop of a conversation. And I’m trying to get it to work on the new circle ci now.
  • a

    adventurous-art-72916

    05/03/2017, 8:47 PM
    haha
  • a

    adventurous-art-72916

    05/03/2017, 8:47 PM
    thanks @rough-sundown-61224 - setting that worked
1...172317241725...2420Latest