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

    cold-journalist-66984

    01/22/2020, 5:26 PM
    I’m seeing a similar issue with zipp, but I’m on pants 1.16. But when I add
    Copy code
    [pytest]
pytest_plugins: +["zipp==1.0.0"]
    I see this error 
    17:18:53 [ERROR] Invalid option 'pytest_plugins' under [pytest]
    What is the right way to pin zipp for pants 1.16?
    w
    h
    • 3
    • 10
  • f

    fancy-queen-20734

    01/22/2020, 6:12 PM
  • c

    calm-artist-46894

    01/23/2020, 5:38 AM
    The default value for 
    --build-file-imports
    is 
    warn
    , what’s the reason to discourage using import in BUILD file? If I need read the environment variable (e.g. the git branch) and do different things based on if the branch is master. Something like:
    Copy code
    def get_version(version):
  import os
  release = os.getenv("CIRCLE_BRANCH") == "master"
  if not release:
    version += ".dev"
  return version
    What is the recommended alternative?
    a
    • 2
    • 3
  • h

    happy-journalist-13561

    01/23/2020, 2:34 PM
    I’m running into an issue when bundling my Javascript webapp. It is not picking up changes I made to any JS files and seems to be using the old version of the artifact. Is there anyway I can validate that changes I make to a Javascript file will be picked up when running node bundle and overwrite the affected cached artifact? The binary is structured in that it builds a JAR for our Play API application and includes the JS bundle with it. I tested this by making a change to one of the JS file and when rebuilding the JAR, it says 1 invalidated target which is what I want. However, when I “unzip” the JAR it does not show the changes I made to that JS file. In addition, when I run the JAR and load the web application, it does not reflect the code logic change I made. For all node build tasks, it is configured to use the Pants contribution package (pants.contrib.node.tasks.node_build...) and I am on 
    Pants 1.16.0
    . What has worked was bumping the 
    cache_key_gen_version
    which will rebuild the entire JAR with new artifacts; however, the expected behavior is that I can reuse the cached artifact and any changes should only override the affected artifacts.
  • f

    fancy-queen-20734

    01/23/2020, 6:41 PM
    could I get a review on https://github.com/pantsbuild/pants/pull/9003 modulo CI et al
    h
    w
    • 3
    • 33
  • w

    witty-crayon-22786

    01/23/2020, 6:44 PM
    @happy-journalist-13561: unfortunately this is a known issue: sorry for the trouble. see https://github.com/pantsbuild/pants/pull/8398
  • w

    witty-crayon-22786

    01/23/2020, 6:45 PM
    if you're able to give that a comment bump, it would be appreciated.
    👍 1
  • e

    early-needle-54791

    01/24/2020, 8:48 PM
    Can you completely suppress pants output so you only get whatever the running target produces as stdout, --quiet doesnt quite do this?
    w
    • 2
    • 12
  • c

    calm-balloon-91412

    01/26/2020, 9:28 PM
    Hi, is there a way to specify 3rdparty libs, some python2 and some python3? also, when pants repl.py 3rdparty/python:somelib -> It opens python2 shell, how can I make it open a python3 shell?
    w
    • 2
    • 7
  • d

    dry-analyst-73584

    01/29/2020, 1:12 AM
    CI seems to be running on a very limited number of shards still. I just had a build take nearly two hours to fail every shard in less than two minutes with an import error (which I cannot repro locally.)
    h
    • 2
    • 5
  • a

    aloof-angle-91616

    01/29/2020, 7:07 AM
    this issue has a few thoughts about whether any utilities for constructing rich/colorful error messages might be useful. it refers to the v2 engine stacktrace as one use case, but if anyone has seen any libraries or tools that have nice error messages, i'd love to draw from any relevant inspiration -- rustc is a great example, they have a cute little framework for formatting compile errors which is mainly why i think an 
    errutil.py
    would ever make sense for pants. https://github.com/pantsbuild/pants/issues/9033
    • 1
    • 4
  • w

    wonderful-iron-54019

    01/29/2020, 1:43 PM
    hey folks wondering if you can help me out here, i have a BUILD file, that contains the following (databricks_job is a custom target) and im getting an ambiguous owner error
    Copy code
    FAILURE: Owners for Resources(BuildFileAddress(src/adobe_mta/BUILD, config)) are ambiguous.  Found DatabricksJob(BuildFileAddress(src/adobe_mta/BUILD, job)) and 1 others: [PythonLibrary(BuildFileAddress(src/adobe_mta/BUILD, lib))]
    Any advice?
    Untitled
    r
    f
    h
    • 4
    • 20
  • l

    little-salesmen-78240

    01/30/2020, 9:37 AM
    Hi everybody, I'm trying to setup a distributed cache for pants in my org, When I try to build on two separate stations the fingerprints for some of the generated targets differ (the exact same commit is checked out)
  • l

    little-salesmen-78240

    01/30/2020, 9:37 AM
    Could you please advise on how to investigate the cause of the difference?
  • l

    little-salesmen-78240

    01/30/2020, 9:39 AM
    So the target is a jvm one, and considering the path is of the shape:
  • l

    little-salesmen-78240

    01/30/2020, 9:39 AM
    Copy code
    /{fingerprint1}/{target-name}/{fingerprint2}-ResolvedJarAwareFingerprintStrategy.{fingerprint3}_{fingerprint4}.tgz
    a
    • 2
    • 1
  • l

    little-salesmen-78240

    01/30/2020, 9:40 AM
    only fingerprint3 differs between the two stations
  • l

    little-salesmen-78240

    01/30/2020, 4:44 PM
    ok so we finally found out that the issue was in the fact pants used absolute file path for fingerprinting targets
  • l

    little-salesmen-78240

    01/30/2020, 4:45 PM
    fixing the same absolute path for the build root on both stations resulted in 100% cache hits, as wanted
    👍 1
  • h

    hundreds-father-404

    01/30/2020, 4:46 PM
    Awesome, glad you were able to get it working. Thank you for sharing the solution
  • l

    little-salesmen-78240

    01/30/2020, 4:47 PM
    Is it possible to avoid this constraint? is the flag --filedeps-absolute=True related to this? We tried with --filedeps-absolute=False which did not result in all cache hits 😕
  • h

    hundreds-father-404

    01/30/2020, 4:50 PM
    I’m not very familiar with how V1 caching works, but others might know the answer there. 
    --filedeps-absolute
    unfortunately wouldn’t help because that only impacts the output of 
    ./pants filedeps
  • a

    aloof-angle-91616

    01/30/2020, 4:51 PM
    Is it possible to avoid this constraint?
    absolutely! and we are actually planning to phase out v1 
    ./pants filedeps
    in favor of a faster version at some point, which is currently available with 
    ./pants fast-filedeps
  • h

    hundreds-father-404

    01/30/2020, 4:52 PM
    absolutely! and we are actually planning to phase out v1 ./pants filedeps in favor of a faster version at some point, which is currently available with ./pants fast-filedeps
    I don’t think Serge is asking about changing how 
    filedeps
    behaves. Rather, he wants to change how cache fingerprinting works in general and was looking at 
    --filedeps-absolute
    as a possible solution that ended up being irrelevant
    a
    • 2
    • 7
  • l

    little-salesmen-78240

    01/30/2020, 4:55 PM
    Thank you for the explanation
  • l

    little-salesmen-78240

    01/30/2020, 4:56 PM
    @aloof-angle-91616 could you please clarify what filedeps is used for?
    a
    h
    • 3
    • 8
  • a

    aloof-angle-91616

    02/04/2020, 3:58 AM
    is there a reason that 
    Target
    doesn't accept the 
    sources
    kwarg, when it assumes that the 
    payload
    will always have a 
    'sources'
    field? I know that i see every target i've ever seen always creates a payload and adds the appropriate sources -- is there a reason we can't have 
    Target
    do that in its ctor?
  • c

    chilly-carpet-91912

    02/04/2020, 12:52 PM
    Hi everyone! I'm struggling with pants requirements while building multiple targets that have disjoint dependencies. Do anyone know solution to handle this?
    Copy code
    Exception message: Could not satisfy all requirements for Markdown==2.1.1:
    Markdown==2.1.1(from: pantsbuild.pants==1.23.0), markdown>=2.6.8(from: tensorflow==1.12.0->tensorboard<1.13.0,>=1.12.0)
    I'm out of ideas
    h
    h
    • 3
    • 33
  • h

    hundreds-breakfast-49010

    02/04/2020, 9:29 PM
    do we currently expect 
    ./pants repl <python_target>
    to populate the repl's namespace with symbols define in the specified target?
    a
    w
    +2
    • 5
    • 44
  • c

    calm-artist-46894

    02/05/2020, 6:24 PM
    Hi what’s the plan to update to 
    pex  2.x
    for pantsbuild? There are more and more libraries dropping python2 support for their latest version and pants will resolve to the latest version for many transitive dependencies (which is not pinned in BUILD) even for python2 target, then we get errors like:
    Copy code
    Exception message: Package SourcePackage('file:///home/bpierce/github.robot.car/cruise/ark/.pants.d/python-setup/resolved_requirements/CPython-2.7.17/grpcio-1.27.0.tar.gz') is not translateable by ChainedTranslator(WheelTranslator, EggTranslator, SourceTranslator)
    We can pin the transitive dependencies in our BUILD file to workaround the issue for now.
    w
    h
    • 3
    • 13
1...213521362137...2420Latest