should i be able to run all tests with

./pants test ::

on pants itself after i cloned it from github main? atm some tests fail.

hi all, maybe a stupid question, but I’m using

experimental_shell_command

to execute maven to build a jar, works perfectly. then, that jar needs to go inside of a

pex_binary

(because I then need my python code to upload it somewhere on demand). pants says I should use a

resource

for that. But the

resource

target only accepts source files. Is there any way I can make the outputs of a previous target into the inputs of the

resources

?

First time using pyoxidizer functionality so excuse the possibly dumb question. I'm trying to setup a very simple example with flask. Repo here: https://github.com/SergeBakharev/pants-example/ It packages okay, but I get a traceback (attached) when running which leads me to believe that flask requires the flaskapp to be on the filesystem and not in memory. I see the

filesystem_resources

option on the

pyoxidizer_binary

target but I can't get it to store a first-party module on the file system. How should I be referencing it?

I want to install a package from the subdirectory of the repo because of its none-standard structure. With pip I can do:

pip install dgllife@ <https://github.com/awslabs/dgl-lifesci/archive/54649fb.zip#subdirectory=python>

But if I put

dgllife@ <https://github.com/awslabs/dgl-lifesci/archive/54649fb.zip#subdirectory=python>

in requirements.txt, pants doesn't seem to recognize options after

#

. Is there a way to make it work?

On Pants 2.14, I'm noticing some new "Pants cannot infer owners for the following imports" warnings. One is for a dependency in a Python test file, which is included in my

[pytest].extra_requirements

list, and another is for a type that is included in

[mypy].extra_type_stubs

, whose import is gated behind an

if TYPE_CHECKING

guard in the test file. This worked without issue in 2.13; is this a bug, or a legitimate new change in behavior? If the latter, what would be the recommended way to fix this? Thanks.

Hi. I’m wondering if pants works with pdm? I did a search in the slack for it and it seem like it at least kind of does but the documentation doesn’t mention poetry or pdm. However, there’s also this blog post which says that it works with poetry. So what’s up with all this?

Hi, soliciting for your dependency hygiene feature requirements (or visibility feature, as it’s been called in the ticket on GH). Please comment on https://github.com/pantsbuild/pants/discussions/17389 with your use cases for what would be desirable in your projects! so that we can make sure it’ll be covered. For example, if you want to limit what part may depend/use some other part in your monorepo, this is the ticket for you.

pants is issuing warnings about "cannot infer owners" (pants 2.14 changed the

[python-infer].unowned_dependency_behavior

default from ignore to warn). But, in at least some of the cases, the indicated lines already have

# pylint: disable=import-error

, so how should I add

# pants: no-infer-dep

when there is already a disable comment for other tools?

Why is pant having issues inferring this

.api_models

import (I'm getting "Pants cannot infer owner" warning): https://github.com/st2sandbox/st2/blob/pants/pylint_plugins/api_models_test.py#L25 Here

/pylint_plugins

is a source root, and

/pylint_plugins/api_models_test.py

is using a relative import for

/pylint_plugins/api_models.py

:

from . import api_models

Does the dep inference not like relative imports?

Can I somehow run an external go dependency? E.g. with python I can make a working

pex_binary

from a dep+name of entrypoint in the dep, but it doesn't seem like I can run a

go_third_party_package

. Context: Want to https://github.com/kubernetes-sigs/controller-tools to generate code.

Hi - I'm trying out the new environments feature in 2.15.0a0 and I'm running into the following issue:

raise InvalidFieldException(
pants.engine.target.InvalidFieldException: Unrecognized field `environment=local_docker` in target src/database:funstuff. Valid fields for the target type `pex_binary`: ['complete_platforms', 'dependencies', 'description', 'emit_warnings', 'entry_point', 'execution_mode', 'ignore_errors', 'include_requirements', 'include_sources', 'include_tools', 'inherit_path', 'interpreter_constraints', 'layout', 'output_path', 'platforms', 'resolve', 'resolve_local_platforms', 'restartable', 'script', 'shebang', 'strip_pex_env', 'tags', 'venv_site_packages_copies'].

I just want to confirm that I have 2.15.0a0 set up properly, since it looks like my pex_binary target isn't accepting

environment

as a field: • I set

pants_version

to 2.15.0a0 • In pants.toml, I set

[environments-preview.names]

to

local_docker="//:local_docker"

• In my topmost BUILD file I added:

docker_environment(
    name="local_docker",
    platform="linux_x86_64",
    image="python:3.9-bullseye",
)

• Lastly, in src/database/BUILD I added:

pex_binary(
    name="funstuff",
    entry_point="funstuff.py",
    environment="local_docker",
)

docker_image(
    name="docker_image",
    instructions=[
        "FROM python:3.9-bullseye",
        'ENTRYPOINT ["/main"]',
        "COPY examples/main.pex /main",
    ],
)

(funstuff.py is just a hello-world Python script). Definitely understand that this is an alpha build but I figured this is likely a basic error w/ something I'm msising. Help would be appreciated - thank you!

Hi, when running a fastapi application using

./pants run <path_to_main.py>

I get this error

ERROR:    Error loading ASGI app. Could not import module "main".

My

main.py

looks like

app = FastAPI(title=PROJECT_NAME, debug=DEBUG, version="0.1.0")
app.include_router(v1_router, prefix="/v1")

if __name__ == "__main__":
    uvicorn.run("main:app", host="0.0.0.0", port=9000, reload=True)

This is because fastapi looks for app in the current directory as mentioned here. Even after providing the complete path from pants, it still complains. The only thing which works is when I modify “main:app” to just

app

and remove the reload option completely. It won’t allow just using

app

with

reload=True

app = FastAPI(title=PROJECT_NAME, debug=DEBUG, version="0.1.0")
app.include_router(v1_router, prefix="/v1")

if __name__ == "__main__":
    uvicorn.run(app, host="0.0.0.0", port=9000, reload=True)

📣 Reminder: the team is currently exploring whether to expand use of GitHub Discussions. Here's background on why you might get asked to move a question over there, or might get asked to post a brief rundown there of what you learned from a particular conversation on slack. (TL;DR: the request is because what you asked was such a good question that we want to ensure those answers are easier for others to discover too!) We apologize for the temporary inconvenience as we evaluate how to take advantage of Discussions to maintain Pants' community's exceptional support. As always, we invite your feedback! If you have experience with GitHub Discussions, whether ours or some other repo's, let us know your POV on the pros and cons of using it for design discussions and/or community tech support. Chime in on either slack or GH.

👋 — how can i debug

ExternalTool

download details? .. i’ve looked at the documentation and according to my own local tools i have entered the proper values for checksum + size, but … pants does not really like me: all i get is (with debug logging as well):

14:51:39.58 [DEBUG] (workunit_store) Completed: Downloading: <https://github.com/google/go-jsonnet/releases/download/v0.19.1/go-jsonnet_0.19.1_Darwin_arm64.tar.gz> (2.87 MB)

14:51:39.58 [ERROR] (pants.bin.local_pants_runner) 1 Exception encountered:

  Exception: Error hashing/capturing URL fetch response: Downloaded file was larger than expected digest

output from running checks on my local system:

❯ shasum -a 256 ~/Downloads/go-jsonnet_0.19.1_Darwin_arm64.tar.gz
c7a98187b4b167e9e885dba52138ed802d4db4635b47b95af6a82448ff9d7ba1  /Users/dreuss/Downloads/go-jsonnet_0.19.1_Darwin_arm64.tar.gz

❯ wc -c /Users/dreuss/Downloads/go-jsonnet_0.19.1_Darwin_arm64.tar.gz
 3004958 /Users/dreuss/Downloads/go-jsonnet_0.19.1_Darwin_arm64.tar.gz

external tool known versions config:

default_known_versions = [
        "v0.19.1|macos_arm64|c7a98187b4b167e9e885dba52138ed802d4db4635b47b95af6a82448ff9d7ba1|3004948",
    ]

how do i get from a given target, to an digest from all its sources + dependencies? … i can resolve a

TransitiveTargetsRequest

which has a

closure

of all

Target

types … but, what is the easiest way to get the full digest to pass to a

Process

from that?

Hello - I am dealing with a pesky third party dependency. Only a source distribution is provided.

setup.py

script imports

numpy

which is not not specified either using

setup_requires

or

PEP-518 build-system requires

. Is there any pants mechanism to patch this type of undeclared dependency? https://www.pantsbuild.org/docs/python-third-party-dependencies#requirements-with-undeclared-dependencies addresses a similar issue, but not for build-dependencies.

when can

ensurepip

step in

export

ever fail? I'm getting error when exporting to venv, just says ensurepip ended with non zero exit. When I check venv out I actually see there's pip module missing so there's that. It doesn't give any other useful error message to check what's going wrong. I can do

pants repl

just fine.

Hi all! Just recently bumped up my project to use pants 2.14.0 from 2.13.0 and I’m trying to use the

xdist_enabled

option. I’ve set it to be true in

pants.toml

, but I’m not exactly sure if my tests are being run in parallel underneath. Checking

htop

isn’t really giving me an indication

If you feel like giving the dep inference blog post some upvote love on the orange site: https://news.ycombinator.com/item?id=33426361

Hi guys. Quick question. We are trying to run django migrations inside of the pex binary, but the migration files end up being generated inside the sandbox, which means they go away. To get around this, we have been running with

-no-run-cleanup

to preserve the files, but it is tedious to copy the files out of the pex sandbox. Anybody dealt with this and have a good solution?

Hi everyone - has anyone tried to run tests for a Django app inside of a Pants monorepo? Curious about other folks' experience with it - if they had to make any BUILD file changes, set dependencies on targets, set up a local database, etc.

something I've noticed. The n`uke_if_too_big` command on the CI/CD page: • Doesn't pass shellcheck • the .cache/pants/setup dir is bigger than 256MB as of 2.13 on a linux system did a PR: https://github.com/pantsbuild/pants/pull/17429

how can i dynamically add a dependency to a target based on a particular fields value? .. i may add that these targets are generated by a

TargetFilesGenerator

… and i’m looking at

settings_request_cls

— but after digging a little bit i’ve not found what i need — i think i need to do something like this … but there’s something i’m not fully understanding yet here i think in terms of how the sausage is made …

hey folks, anyone had success pre-compiling Python code to .pyc and removing .py on Python targets ?

Hi everyone - we are considering Pants for a project where we are moving from a monorepo to a "framework + plugin components" model, such that the plugins are versioned/packaged discretely. In our current POC we're using namespace packages for the plugins, and I had hoped to use pex directly to deploy each plugin in an isolated environment, but I don't believe pex files can be run on Windows. Does anyone see a path forward for this scenario with Pants?

okay, so i’m using vscode for pants plugin development .. and was wondering if anyone has some tips on language server/ide setup? — i think i got most of stuff working, but if im in a module where i have not yet imported

Process

or

ProcessResult

— i’m not being able to auto import and manage the imports for it … — while if i reference symbols from my local plugin it works just fine — when i do import by hand, vscode knows where the source can be found and i can jump directly into definitions … anyway to make that include detection smarter?

Is there a way to prevent our pex-controlled lockfiles from locking deps for platforms we don't support (ie. i686, aarch64). We exclusively use x86_64, but I'm not seeing options that would let us restrict to that arch

Has anyone run into issues with DistributionNotFound error with the google cloud functions integration?

Hi all! Our CI seems to have recently broken during the

./pants tailor --check

run. We haven’t seem this happen before but we seem to be hitting:

No valid Python interpreter found. For `pants_version = "2.13.0"`, Pants requires Python 3.7, 3.8, or 3.9 to run. Please check that a valid interpreter is installed and on your $PATH.

HI everyone, I’m trying out Pants after listening to the Talk Python To Me Podcast. 🙂