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

    bored-art-40741

    08/30/2015, 6:05 PM
    I was going to do some work today on having jvm missing dep detection consume from 
    compile_classpath
    and eliminate its dependency on 
    ivy_jar_products
    , but I don't want to fiddle with code that's going to become a big merge conflict
  • b

    bored-art-40741

    08/30/2015, 6:34 PM
    Posted an RB with the revert. @adventurous-twilight-44110 it doesn't seem like I can add you as a reviewer, but I added Stu and I'm pinging you here so you can take a look: https://rbcommons.com/s/twitter/r/2728/
  • w

    witty-crayon-22786

    08/30/2015, 11:26 PM
    @bored-art-40741: yes, the products that power missing dep detection are stored in the cache.
  • w

    witty-crayon-22786

    08/30/2015, 11:26 PM
    it's the analysis file.
  • w

    witty-crayon-22786

    08/30/2015, 11:27 PM
    is there a reproduction for it not firing the second time?
  • b

    bored-art-40741

    08/30/2015, 11:27 PM
    No, I didn't attempt to repro, I was basing this on the conversion in the RB comments
  • w

    witty-crayon-22786

    08/30/2015, 11:28 PM
    i think you might have the wrong idea then
  • w

    witty-crayon-22786

    08/30/2015, 11:28 PM
    the dep check task will fail, continuously
  • w

    witty-crayon-22786

    08/30/2015, 11:28 PM
    until the missing dep is declared
  • w

    witty-crayon-22786

    08/30/2015, 11:28 PM
    and since the dep check task is installed in 
    compile
    , compile will not succeed
  • b

    bored-art-40741

    08/30/2015, 11:28 PM
    Hm
  • w

    witty-crayon-22786

    08/30/2015, 11:28 PM
    the compile.jvm task(s) will succeed,
  • w

    witty-crayon-22786

    08/30/2015, 11:28 PM
    but the goal will fail
  • b

    bored-art-40741

    08/30/2015, 11:28 PM
    Do we have a test that demonstrates this?
  • b

    bored-art-40741

    08/30/2015, 11:28 PM
    That's great if it's the case
  • b

    bored-art-40741

    08/30/2015, 11:29 PM
    I didn't expect the analysis to be reloaded into products even when the input target is fully cached
  • w

    witty-crayon-22786

    08/30/2015, 11:29 PM
    that a missing dep fails compilation?
  • w

    witty-crayon-22786

    08/30/2015, 11:29 PM
    isn't that what all the tests do?
  • w

    witty-crayon-22786

    08/30/2015, 11:29 PM
    yes, the analysis is reloaded.
  • b

    bored-art-40741

    08/30/2015, 11:31 PM
    I actually had trouble finding any tests of this
  • b

    bored-art-40741

    08/30/2015, 11:31 PM
    Can you point me at them?
  • b

    bored-art-40741

    08/30/2015, 11:31 PM
    I want to write some more to support the change from 
    ivy_jar_products
    to 
    compile_classpath
  • w

    witty-crayon-22786

    08/30/2015, 11:32 PM
    tests/python/pants_test//backend/jvm/tasks/jvm_compile/java/test_java_compile_integration.py
  • w

    witty-crayon-22786

    08/30/2015, 11:32 PM
    tests/python/pants_test//backend/jvm/tasks/jvm_compile/java/test_java_compile_integration.py
  • b

    bored-art-40741

    08/30/2015, 11:33 PM
    Yes, I saw the whitelist test
  • b

    bored-art-40741

    08/30/2015, 11:33 PM
    Okay, I guess that is in fact the only test of this feature
  • b

    bored-art-40741

    08/30/2015, 11:34 PM
    It doesn't check artifact cache semantics though, and it's primarily geared toward making sure that the whitelist works rather than missing dep detection
  • b

    bored-art-40741

    08/30/2015, 11:34 PM
    So no, the existing tests don't do the test we need here, if this is the only test
  • w

    witty-crayon-22786

    08/30/2015, 11:36 PM
    it tests both positives and negative cases twice
  • w

    witty-crayon-22786

    08/30/2015, 11:36 PM
    so a re-run will re-fail.
1...241242243...2420Latest