Will report back momentarily

So, /mnt/c is made available to the Linux VM via a plan9 network filesystem that does not support file watching.

I have no clue why they did not use Samba / SMB, very very odd. But that would likely have same issue .

Is anyone having success/seeing benefit with managing a repo of Airflow DAGs with Pants? Is there anything to look out for?

i often want to change my filesystem while running pytest, but I get `Filesystem changed during run: retrying

Test

in 500ms...` . I’m pretty confident my changes shouldn’t affect the test (And that in this case I’m not even changing any transitive dependencies of the test). is there a flag I can pass to

pants test

to ignore filesystem changes during run?

At the moment,

check

fails for me on GitHub but works locally. The strange thing is that it complains that it can't download some file, but the link works. I trie to rerun the job multiple times... Any idea please? 🙂

Hi, has anyone set-up dependabot with pants? How would it work? I see it works without much configuration with poetry lock files. Not sure how it would work with pex generated lockfiles I don’t see any example in pants own repo either. Currently it looks like pants itself has only configured it for rust. https://github.com/pantsbuild/pants/blob/main/.github/dependabot.yml

How do I set up pants so that I can use

experimental_shell_command

? I've set up a target and I'm getting

name 'experimental_shell_command' is not defined

I don't see anything here: https://www.pantsbuild.org/docs/run-shell-commands adding

pants.backend.experimental_shell_command

to

backend_packages

doesn't work. Pretty sure this has got to be a silly little configuration thing I haven't grasped yet

Aha, it's

pants.backend.shell

Follow-up question: Where does the command look for my script file?

experimental_shell_command(
    command="./script.sh",
    tools=["echo", "bash"],
    dependencies=[":docker"],
)

Same folder as the BUILD file?

it doesn't seem to run wherever I put it, but I could be doing something else wrong 🤷‍♂️

Hi all! I have got a kind of a FAQ problem all over again, but no known solutions are helping... So I have a Pants project that I had on hold, and now got back to it. It had been working before, but now it doesn't anymore. I have a Python script that executes a couple of pants commands, to check some dependencies. Running it manually works ok, but running it via "PANTS_CONCURRENT=True ./pants run " doesn't, I get "Failed to begin watching the filesystem: Too many open files (os error 24)". This is with pants 2.14.0. Has anyone any suggestions about why it doesn't work anymore? Should I just find another way to run the script?

Are there still limitations with using Python versions >3.9 with Pants?

When using the new 2.15 pants binary, what should be the procedure in CI? Install with pantsup.sh every time, on the CI machine? The actions/init-pants action does some fancy initialization for github workflows, is there any info for generic CI?

Hi! I'm having a problem similar to https://pantsbuild.slack.com/archives/C046T6T9U/p1673569181883479. In GitHub Actions, I can successfully log in to Azure Container Registry, then do

docker pull

. But if

pants test

(version 2.15.0rc2) tries to do the same, it results in:

Exception: Failed to pull image `***<http://registry.azurecr.io/runner/ubuntu20_***|registry.azurecr.io/runner/ubuntu20_***>`: Failed to pull Docker image `***<http://registry.azurecr.io/runner/ubuntu20_***|registry.azurecr.io/runner/ubuntu20_***>`: DockerResponseServerError { status_code: 500, message: "Get \"[https://***<http://registry.azurecr.io/v2/runner/ubuntu20_***/tags/list\|registry.azurecr.io/v2/runner/ubuntu20_***/tags/list\>](https://***<http://registry.azurecr.io/v2/runner/ubuntu20_***/tags/list/|registry.azurecr.io/v2/runner/ubuntu20_***/tags/list/>)": unauthorized: authentication required, visit <https://aka.ms/acr/authorization> for more information." }

I can see that the ACR login steps yields this:

DOCKER_CONFIG=/home/runner/work/_temp/docker_login_1675410905572

I have tried with:

[docker]
build_args = ["GIT_SHA"]
env_vars = ["DOCKER_CONFIG"]

[test]
extra_env_vars = ["DOCKER_CONFIG"]

as well as without the

env_vars

and

extra_env_vars

above, but it fails either way.

Hi! I'm trying to figure out if it's possible to get pants to use a .netrc file for a private pypi repo? We currently have e.g. a .pants.rc to specify the repo and auth, which works fine but it would be nice to have a single place to refer to (since e.g. other projects use pip direct) for auth, and include the repo path without the auth within the pants.toml

Is there a way to build a pex for commands (such as

pants test

or

pants run

) without trying to bundle 3rd party code? I know you can use PEX_INHERIT_PATH=prefer when running the pex but I'm not sure how to tell Pants to not even try to resolve the 3rd party deps to put them in the pex to begin with.

No. Clearly the 3rd party code is needed though; so what's your plan / problem / scheme?

Is this roughly "Pants / Pex are too slow, just please use my pre-built venv with all deps for all code built in? - trust me, this is not hermetic but I have carefully arranged this venv to exist on all machines."

This is "all my 3rd party libs are actually provided by RPMs, not Pip, but I lied to Pants and made up fake python packages for all of them so I could turn on `unowned_dependency_behavior = error`"

Maybe I could make a fake resolve that excludes my fake deps, (don't think this will work) and then pass

--unowned-depdency-behavior=ignore

when running

pants test

.

I'm aware this is all kinda insane and half-cocked

Yeah, I think the two options are continuing down the Rube Goldberg path, but that requires Pants mods to support; so that's a further sign, or working out 1st class support for this sort of scenario.

Is there a way to force a particular manylinux when building a pex? The pex is built on a newer glibc than the one we run it on

I mean, the "eventual" fix would be first class support of execution of stages in containers and

rpm_python_*

targets that could be contributed back as a backend

I was looking for something hackier to do a proof of concept in the short-term

Yeah, your hacks would need to expand into Pants mods. Pants currently only supports splitting PEXes when running

./pants package

.

So your hack probably needs to involve a Pants branch and using

pants_from_sources

to demo in the short term.

Is there a way to treat warnings as failures? For example I have warnings like:

[WARN] Pants cannot infer owners for the following imports in the target <TARGET>

which lead to failures further on. As this is a warning though, it doesn’t stop the build until a problem occurs later. I would like to fail the build on issues like this so it doesn’t get hidden in the logs