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

    witty-crayon-22786

    03/02/2016, 9:36 PM
    ...and shouldn't, because we can trust it now
  • w

    witty-crayon-22786

    03/02/2016, 9:36 PM
    git clean -fdx
    does
  • w

    witty-crayon-22786

    03/02/2016, 9:37 PM
    but putting this in the buildcache would also make it subject to remote caching, so even that gets lighter.
  • w

    witty-crayon-22786

    03/02/2016, 9:37 PM
    er... and to be clear, the default build cache is in the buildroot, not the workdir
  • a

    ancient-stone-50795

    03/02/2016, 9:39 PM
    OK, we aren't using defaults:
    Copy code
    read_from: [ "%(pants_workdir)s/buildcache" ]
read: True
write_to:  [ "%(pants_workdir)s/buildcache" ]
write: True
  • a

    ancient-stone-50795

    03/02/2016, 9:39 PM
    Should I just nuke these?
  • a

    ancient-stone-50795

    03/02/2016, 9:40 PM
    I think so. I see when I do I get the default you're talking about.
  • w

    witty-crayon-22786

    03/02/2016, 9:44 PM
    @ancient-stone-50795: yea, definitely. we've resolved all known issues with the cache, and it is trustworthy enough to leave in the buildroot. you should confirm the same for any custom tasks you have but
  • w

    witty-crayon-22786

    03/02/2016, 9:44 PM
    i think even @many-engine-77019 would agree =P
  • w

    witty-crayon-22786

    03/02/2016, 9:44 PM
    ...but i'm not holding my breath
  • m

    many-engine-77019

    03/02/2016, 9:45 PM
    didn't you see me defending that on finaglers a couple weeks ago?
    metal 1
  • w

    witty-crayon-22786

    03/02/2016, 9:45 PM
    aw shucks
  • b

    bored-art-40741

    03/02/2016, 10:12 PM
    So the stack breakage in CI is not due to concurrency
  • b

    bored-art-40741

    03/02/2016, 10:12 PM
    It happens when there's a single shard running
  • b

    bored-art-40741

    03/02/2016, 10:12 PM
    I'm pretty confident this is the bad code
  • b

    bored-art-40741

    03/02/2016, 10:12 PM
    Copy code
    @memoized_property
  def base_dir(self):
    """Returns the base directory for this stack distribution.

    :returns: The stack distribution base directory.
    :rtype: string
    """
    stack_dist = self._binary_util.select_binary(self._relpath, self.version, 'stack.tar.gz')
    distribution_workdir = os.path.dirname(stack_dist)
    outdir = os.path.join(distribution_workdir, 'unpacked')
    if not os.path.exists(outdir):
      with temporary_dir(root_dir=distribution_workdir) as tmp_dist:
        TGZ.extract(stack_dist, tmp_dist)
        os.rename(tmp_dist, outdir)
    return os.path.join(outdir, 'stack')
  • b

    bored-art-40741

    03/02/2016, 10:12 PM
    It's unpacking a thing under the directory that select_binary gave back to it
  • b

    bored-art-40741

    03/02/2016, 10:13 PM
    Which I don't think we make any guarantees about
  • b

    bored-art-40741

    03/02/2016, 10:20 PM
    I don't think any of our other uses of select_binary try to be clever
  • b

    bored-art-40741

    03/02/2016, 10:20 PM
    They just assume they've been handed back an executable
  • b

    bored-art-40741

    03/02/2016, 10:42 PM
    Travis is getting ridiculous
  • b

    bored-art-40741

    03/02/2016, 10:42 PM
    Hours to get a run through
  • l

    limited-country-20626

    03/02/2016, 11:14 PM
    @red-cpu-25099: Patrick and I are both failing Travis runs with the following error: https://travis-ci.org/pantsbuild/pants/jobs/113242203. I saw that your pull request passed - but do you have any thoughts before I dig?
  • f

    full-kilobyte-93429

    03/02/2016, 11:25 PM
    he is in London... looks like from https://rbcommons.com/s/twitter/r/3527 not familiar with the 
    spec_excludes
    deprecation, but since it says
  • f

    full-kilobyte-93429

    03/02/2016, 11:25 PM
    Copy code
    'Filtering broken BUILD files based on exclude_target_regexp is deprecated '
  • f

    full-kilobyte-93429

    03/02/2016, 11:25 PM
    the broken test can probably be removed 
    test_exclude_target_regexps
  • l

    limited-country-20626

    03/02/2016, 11:57 PM
    yeah that is it. master branch is broken.
  • l

    limited-country-20626

    03/02/2016, 11:59 PM
    I just sent up a few monster code reviews that move the android backend into contrib, I tried to break it up in a way that made review at least possible. https://rbcommons.com/s/twitter/r/3526/ gets the android backend green and fixes the example projects. CR 3530 does the mostly mechanical creation of the android.contrib module. CR 3531 removes any special casing that was patched into core to handle the fact that android was it's own backend.
  • a

    ancient-stone-50795

    03/03/2016, 12:00 AM
    @jolly-chef-92794: reported that he tried running the Square build with the latest patns master including the change to ivy from Stu and it broke.
  • l

    limited-country-20626

    03/03/2016, 12:00 AM
    My goal was solely to get android out of core. I got the android backend tests and examples green but by force.
1...102710281029...2420Latest