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

    witty-crayon-22786

    02/15/2017, 8:03 PM
    but maybe just parsing it for now would be enough, as using fetcher would probably make swapping it to v2 harder
  • f

    full-kilobyte-93429

    02/15/2017, 8:05 PM
    not sure i get the v2 part
  • w

    witty-crayon-22786

    02/15/2017, 8:06 PM
    @full-kilobyte-93429 : v2 will need native support for fetching things
  • w

    witty-crayon-22786

    02/15/2017, 8:42 PM
    @happy-kitchen-89482 : so, you'll be low on context, but: if you explicitly pass a 
    FingerprintStrategy
    to the 
    Task.invalidated
    block, then you override the default (
    TaskIdentityFingerprintStrategy
    ), which includes a bunch of super relevant information including options and tool versions.
  • w

    witty-crayon-22786

    02/15/2017, 8:43 PM
    i'm thinking of basically requiring 
    TaskIdentityFingerprintStrategy
    , and having it delegate to whatever the custom strategy is... ie
    Copy code
    fingerprint_strategy = TaskIdentityFingerprintStrategy(self, delegatee=fingerprint_strategy)
    in 
    Task.invalidated
  • w

    witty-crayon-22786

    02/15/2017, 8:44 PM
    alternatively, moving the information in 
    TaskIdentityFingerprintStrategy
    down a level to the CacheManager or something.
  • w

    witty-crayon-22786

    02/15/2017, 8:44 PM
    but either way: making it impossible to disable including option fingerprints in a task's fingerprints
  • w

    witty-crayon-22786

    02/15/2017, 8:45 PM
    (sidenote: can't wait to nuke this.)
  • w

    witty-crayon-22786

    02/15/2017, 8:47 PM
    the question is basically: can you think of a case where a task shouldn't mix in options that have been marked 
    fingerprint=True
    ?
  • w

    witty-crayon-22786

    02/15/2017, 8:47 PM
    now that i'm typing it out, the answer seems to be "obviously not"
  • w

    witty-crayon-22786

    02/15/2017, 8:50 PM
    (for color: i failed to notice while reviewing https://github.com/pantsbuild/pants/pull/4177, that explicitly specifying a strategy was nuking the good defaults of 
    TaskIdentityFingerprintStrategy
    , meaning that the fingerprint of the scrooge jar was no longer accounted for)
  • h

    happy-kitchen-89482

    02/15/2017, 8:50 PM
    Yeah, although 
    delegatee
    is just called a 
    delegate
    🙂
  • h

    happy-kitchen-89482

    02/15/2017, 8:50 PM
    Also this can be more straightforwardly done with subclassing
  • w

    witty-crayon-22786

    02/15/2017, 8:50 PM
    it can, but if you forget, you get in trouble
  • w

    witty-crayon-22786

    02/15/2017, 8:50 PM
    as here
  • h

    happy-kitchen-89482

    02/15/2017, 8:50 PM
    I.e., you have to extend 
    TaskIdentityFingerprintStrategy
    , and implement an 
    extra_fingerprint_ingredients
    method or something,
  • w

    witty-crayon-22786

    02/15/2017, 8:51 PM
    so my question is: are there valid reasons to not want to include this info
  • h

    happy-kitchen-89482

    02/15/2017, 8:51 PM
    None that I can think of off-hand
  • w

    witty-crayon-22786

    02/15/2017, 8:52 PM
    i think we might have been afraid of options fingerprints up front? but they've proven to be insanely useful as a default
  • w

    witty-crayon-22786

    02/15/2017, 8:52 PM
    so yea... i think i'll push the task fingerprint down to the CacheManager to make them impossible to disable, and deprecate 
    TaskIdentityFingerprintStrategy
  • w

    witty-crayon-22786

    02/15/2017, 8:52 PM
    (as it would be redundant at that point)
  • w

    witty-crayon-22786

    02/15/2017, 9:02 PM
    @happy-kitchen-89482 : thanks!
  • w

    witty-crayon-22786

    02/15/2017, 9:33 PM
  • w

    witty-crayon-22786

    02/15/2017, 9:37 PM
    @happy-kitchen-89482 : thoughts on https://github.com/pantsbuild/pants/issues/2901 ?
  • m

    miniature-nightfall-50290

    02/15/2017, 11:23 PM
    qq how do I enable the built-in cobertura coverage plugin?
  • m

    miniature-nightfall-50290

    02/15/2017, 11:23 PM
    didn’t see anything obvious, and it does’t looke like it’s on pypi
  • w

    wide-energy-11069

    02/15/2017, 11:26 PM
    Copy code
    $ ./pants options | grep cov
pytest.cov_requirements = pytest-cov>=1.8,<1.9 (from HARDCODED)
test.junit.coverage = False (from HARDCODED)
test.junit.coverage_cobertura_exclude_classes = [] (from NONE)
test.junit.coverage_cobertura_include_classes = [] (from NONE)
test.junit.coverage_force = False (from HARDCODED)
test.junit.coverage_jvm_options = [] (from NONE)
test.junit.coverage_open = False (from HARDCODED)
test.junit.coverage_processor = cobertura (from HARDCODED)
test.pytest.coverage = None (from NONE)
test.pytest.coverage_output_dir = None (from NONE)
  • m

    miniature-nightfall-50290

    02/15/2017, 11:27 PM
    oh dope
  • p

    purple-oil-45048

    02/16/2017, 1:08 AM
    is see the 
    test.pytest.junit_xml_dir
    option .. is the xml output also written somewhere by the 
    test.junit
    goal?
  • w

    witty-crayon-22786

    02/16/2017, 1:13 AM
    yep
1...163916401641...2420Latest