https://pantsbuild.org/ logo
Powered by
#development
  • a

    average-vr-56795

    05/10/2019, 8:58 PM
    Yay, futures just released my change! No more forking!
    🥄 2
    🎉 3
    đź’Ż 2
    🔥 2
    đź‘– 2
  • a

    aloof-angle-91616

    05/10/2019, 8:59 PM
    thank you for spending the effort to avoid that possible tech debt!
  • h

    hundreds-father-404

    05/11/2019, 2:36 AM
    Issue with the V2 source root work. Now Pytest will output the relative file names, rather than the full. For example, instead of
    Copy code
    =========================== 1 passed in SOME_TEXT ===========================


testprojects/tests/python/pants/dummies:target_with_thirdparty_dep              .....   SUCCESS
    we get
    Copy code
    =========================== 1 passed in SOME_TEXT ===========================


python/pants/dummies:target_with_thirdparty_dep                      .....   SUCCESS
    w
    • 2
    • 1
  • a

    aloof-angle-91616

    05/11/2019, 2:36 AM
    hm. are the pytests being executed in the buildroot, or in 
    testprojects/
    ?
  • h

    hundreds-father-404

    05/11/2019, 2:37 AM
    The Pex we build has the files with 
    testprojects/tests
    stripped, which we had to do to get the imports to work
  • a

    aloof-angle-91616

    05/11/2019, 2:38 AM
    in zinc compile hermetic, for example, we always execute with cwd=get_buildroot() basically
  • a

    aloof-angle-91616

    05/11/2019, 2:38 AM
    oh
  • a

    aloof-angle-91616

    05/11/2019, 2:38 AM
    create the directory and move it
  • h

    hundreds-father-404

    05/11/2019, 2:38 AM
    I think the execution model is different. Here, we create a Pex with solely the Pants files necessary to get things working
  • a

    aloof-angle-91616

    05/11/2019, 2:39 AM
    argv=tuple(['/bin/sh', '-c', '/bin/mkdir -p testprojects/tests/ && /bin/mv python/ testprojects/tests/ && <my command...>'])
  • a

    aloof-angle-91616

    05/11/2019, 2:40 AM
    we do the same there
  • a

    aloof-angle-91616

    05/11/2019, 2:40 AM
    the thing is we just make it a snapshot instead of bundling it into anything
  • a

    aloof-angle-91616

    05/11/2019, 2:40 AM
    is there a reason we need to make a pex in the first place right now instead of a collection of snapshots
    h
    • 2
    • 1
  • a

    aloof-angle-91616

    05/11/2019, 2:43 AM
    well, yes -- so that the files will be in a reliable place in the execution environment. forget me
  • h

    hundreds-father-404

    05/11/2019, 3:05 AM
    All is not lost. We will still output 
    python/pants/dummies:target_with_source_dep_relative_import stdout:
    . It’s only the actual stdout that won’t have the full name. I definitely think we should leave this as a TODO
    a
    • 2
    • 5
  • a

    aloof-angle-91616

    05/12/2019, 8:04 PM
    just merged https://github.com/pantsbuild/pants/pull/7532 (hadn't had time to investigate the final CI failure until just now when i found it was a flake) so lots of errors should be better. i think this unblocks the first 
    --query
    PR as well since the failure i was seeing was just a previously-hidden failure
    ❤️ 1
  • a

    aloof-angle-91616

    05/12/2019, 9:22 PM
    i also just got python thrift namespace checking to work as a console rule: https://github.com/pantsbuild/pants/pull/7419 going to convert my v1 task testing for that into v2 testing which should be fun
    đź’Ż 1
  • a

    aloof-angle-91616

    05/13/2019, 3:20 AM
    running pants v2 rules from source is such a fast edit/compile/debug loop
  • a

    aloof-angle-91616

    05/13/2019, 3:21 AM
    Copy code
    > time ./pants --enable-pantsd validate-python-thrift examples/src/thrift/org/pantsbuild/example/precipitation:monolithic-precipitation-python
clashing namespaces for python thrift library sources detected in build graph:
org.pantsbuild.example.monolithic_precipitation: [(examples/src/thrift/org/pantsbuild/example/precipitation:monolithic-precipitation-python, examples/src/thrift/org/pantsbuild/example/precipitation/monolithic_precipitation.thrift), (examples/src/thrift/org/pantsbuild/example/precipitation:monolithic-precipitation-python, examples/src/thrift/org/pantsbuild/example/precipitation/other.thrift)]
./pants --enable-pantsd validate-python-thrift   0.87s user 0.37s system 80% cpu 1.541 total
  • a

    aloof-angle-91616

    05/13/2019, 3:21 AM
    i added a broken file to that target and am running the 
    validate-python-thrift
    console_rule
  • a

    aloof-angle-91616

    05/13/2019, 3:26 AM
    (and that's 1.541 seconds, tl;dr)
  • a

    aloof-angle-91616

    05/13/2019, 3:26 AM
    its pretty reliably 2 seconds without pantsd
  • a

    aloof-angle-91616

    05/13/2019, 3:27 AM
    working on testing now
  • a

    average-vr-56795

    05/13/2019, 10:15 AM
    @aloof-angle-91616 Cool! Is the plan to turn the 
    Task
    into a 
    @console_rule
    , or to land it as a 
    Task
    ?
    a
    • 2
    • 3
  • a

    average-vr-56795

    05/13/2019, 12:12 PM
    I wasn’t crazy! Sometimes we were going out of sync between native.py and the FFI signatures! https://github.com/pantsbuild/pants/pull/7708 It was just a very specific set of actions you needed to take to trigger it!
    🔥 1
    a
    • 2
    • 1
  • a

    aloof-angle-91616

    05/13/2019, 3:42 PM
    @witty-crayon-22786 @average-vr-56795 any idea why a sources snapshot would return the empty digest on a target i've created with 
    self.add_to_build_file()
    in a 
    ConsoleRuleTestBase
    ? i'm going to have to table working on this for now, but the 
    @console_rule
    itself works, and the testing should too, i just haven't had time to dive into how to get my file contents into the sources snapshot. https://github.com/pantsbuild/pants/pull/7419
    a
    • 2
    • 14
  • h

    hundreds-father-404

    05/14/2019, 12:43 AM
    With V2 pytest, will it no longer be possible to use PDB in a test because it’s executed as a subprocess?
    🔥 1
    a
    w
    • 3
    • 14
  • a

    aloof-angle-91616

    05/14/2019, 12:43 AM
    we can probably transfer the console, we would just have to write that
    đź’Ż 1
  • a

    aloof-angle-91616

    05/14/2019, 12:43 AM
    it can be as complex as we want it to be
  • h

    hundreds-father-404

    05/14/2019, 12:44 AM
    yay, glad to know it’s a possibility
1...625626627...906Latest