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

    full-kilobyte-93429

    01/11/2017, 10:13 PM
    @witty-crayon-22786 whenever you get a chance, can you do a zinc publish so i will finish the python side? nothing urgent
  • w

    witty-crayon-22786

    01/11/2017, 10:14 PM
    sure
  • b

    billowy-beach-68498

    01/11/2017, 10:41 PM
    @brash-oxygen-26029 last time I tried travis_wait, it weirdly broke some other tests in ci and I don't know the correlation. I filed an issue in travis github page but they don't know the cause either. That's why I split the test and left that comment.
  • b

    billowy-beach-68498

    01/11/2017, 10:42 PM
    but I think you can try that again to see if the error remains.
  • b

    billowy-beach-68498

    01/11/2017, 10:53 PM
    BTW this was the issue I filed back then: https://github.com/travis-ci/travis-ci/issues/6513
  • b

    brash-oxygen-26029

    01/12/2017, 12:46 AM
    travis_wait
    didn't work for me either. The integration tests timed out the same way they were timing out before. Instead I just excluded all testproject/* projects from running Error Prone and the test times came back down under 10 minutes and passed.
  • f

    fast-megabyte-33579

    01/12/2017, 10:53 AM
    This Eta lang looks interesting - http://eta-lang.org/
  • h

    high-advantage-47550

    01/12/2017, 7:25 PM
    I’m getting
    Copy code
    No such file or directory: u'/usr/src/app/.pants.d/compile/zinc/252d64521cf9/myproject.src.main.scala.scala/current/myproject.src.main.scala.scala.analysis'
    but even removing 
    .pants.d
    and re-running isn’t fixing — any sugggestions?
  • h

    high-advantage-47550

    01/12/2017, 7:35 PM
    oh, but removing 
    .cache
    fixed 🙂
  • w

    witty-crayon-22786

    01/12/2017, 9:35 PM
    @high-advantage-47550: hm. that's very suspect. what version?
  • h

    high-advantage-47550

    01/12/2017, 9:35 PM
    1.2.0
  • w

    witty-crayon-22786

    01/12/2017, 9:36 PM
    you might try switching to the 1.2.1 release candidate: 
    1.2.1rc0
  • w

    witty-crayon-22786

    01/12/2017, 9:37 PM
    it will become 1.2.1 whenever i get a chance to lift it
    👍 1
  • h

    high-advantage-47550

    01/12/2017, 9:38 PM
    will do! thanks
  • h

    high-advantage-47550

    01/13/2017, 1:46 AM
    why would 
    ./pants test myproject/src/test/scala:scala
    pass but that same suite fail during 
    ./pants test ::
    ? does the classpath change?
  • w

    witty-crayon-22786

    01/13/2017, 1:48 AM
    @high-advantage-47550: that's my comment from yesterday about 
    ::
    including everything you have defined, and thus potentially resulting in a different ivy resolve
  • w

    witty-crayon-22786

    01/13/2017, 1:48 AM
    the other aspect is that there is a single junit invoke, which is more likely to cause issues
  • h

    high-advantage-47550

    01/13/2017, 1:48 AM
    yeah, that makes sense — thank you
  • w

    witty-crayon-22786

    01/13/2017, 1:49 AM
    invoking junit exactly once is a "feature", but the ability to invoke it once per target would be much appreciated
  • w

    witty-crayon-22786

    01/13/2017, 1:50 AM
    there is an odd middle ground which is to use the 
    --test-junit-batch-size
    option, but it's not very polished: it works at the test level.
  • w

    witty-crayon-22786

    01/13/2017, 1:51 AM
    in CI, we do something like:
    Copy code
    (./pants list :: > targets.txt) && ./pants --target-spec-file=targets.txt minimize
  • w

    witty-crayon-22786

    01/13/2017, 1:52 AM
    and then invoke once per minimized test target
  • h

    high-advantage-47550

    01/13/2017, 1:53 AM
    i was just about to ask! thanks for the insights
  • w

    witty-crayon-22786

    01/13/2017, 1:53 AM
    needing to invoke multiple times is awkward, frankly. but it aligns with what we do in CI because we run the tests across multiple workers/shards
  • f

    fast-megabyte-33579

    01/13/2017, 9:41 AM
    the behaviour is also different between 
    test //::
    and 
    compile //::
    I think - compile is “do this once for each target I find” not “compile everything all together"
  • l

    limited-country-20626

    01/13/2017, 5:43 PM
    yes - that is true. I have several CIs that run 
    ./pants compile src:: test::
    and 
    ./pants test test::
    - we use them to sanity check our test health and prime the cache
  • l

    limited-country-20626

    01/13/2017, 5:43 PM
    but we also don't use ivy
  • h

    high-advantage-47550

    01/13/2017, 7:08 PM
    not that I expect it to be particularly unusual, but I might as well ask: does anyone have a nice setup for integration tests with pants?
  • h

    high-advantage-47550

    01/13/2017, 7:36 PM
    is it possible to specify a dependency on a bundle, specifically?
  • h

    high-advantage-47550

    01/13/2017, 7:37 PM
    i was planning to do 
    pants run
    in a 
    prep_command
    but that won’t work due to the file lock
1...159115921593...2420Latest