Hi guys. Has anybody been able to make a pyright plugin for pants? We are running into an annoying mypy error called

Duplicate module named {xxxx}

and it looks like there isn't much of a workaround for it

Hello, one of our engineers is hitting a strange error when running pants commands, and I haven’t had any luck with debugging. Could anyone by chance point out where I should be looking? (full error in 🧵 )

Any ideas why

./pants lint ::

would succeed locally and then fail on some files on the same branch in CI? This has happened once or twice on my team. Recloning the branch into a new directory and running

./pants fmt ::

properly catches the files that need reformatting. Could it be a local cache issue?

Prior to

2.13.0

I was using a common Dockerfile for multiple targets and referring to it with a relative path:

docker_image(
    name="docker",
    source="../Dockerfile.prod",
    ...
)

After upgrading to

2.13.0

I am now getting:

Copy code

InvalidFieldException: The 'source' field in target xxx:docker should not include `../` patterns because targets can only have sources in the current directory or subdirectories

I have tried creating a

files

target and including it into

docker_image.dependencies

but am not having any luck with that either Is there another way I should try? Should I open an issue to track on GH?

Is there a way how, within a CI, to trigger a build of docker images whose dependencies have changed? I mean - I would like my CI to build an image if Python sources that are going to be copied into it (as a PEX file) or a Dockerfile were changed. Something like

./pants --changed-since=origin/master --changed-dependees=transitive test

but for building images 🙂

Hi, does pants run

go generate

before compiling?

Hi all! I'm testing Pants with our Python project and we have an internal pypi repository that needs authentication to access. I don't want to add the credentials to pants.toml, of course, so the question is if there is a way to refer to environment variables from there? Or do I have to send the repo URL on the command line every time?

I created a couple of dummy projects in a repo, and trying to see how the multiple lockfiles feature works. Question: do I need to use resolve="something" in each and every BUILD file in the projects? I could not find a way to specify it for the whole project, and I get errors if it's not everywhere.

Also, I sometimes (maybe every 3 times I run a command) an error saying "unmatched globs in project's source fields" but there are no changes in any file between the runs, so I am confused. Do I miss some config?

Hi, I have added a timeout to the python tests and when a test fails, it doesn’t print any logs of the test. For example, when during a test an exception is raised, then I see logs of the test. Do I miss any additional configuration related to the timeout or is it not handled? Thank you!

Exceeded timeout of 600.0 seconds when executing local process: Run Pytest for …

I am running

pants generate-lockfiles

and it just sits there counting up the time, it's at over 5 minutes now. How can I see what the problem is?

-ldebug

shows that the pex command is started, but nothing more than that....

I think that message comes from the daemon, when it starts.

./pants test --test-timeout-default=20  --test-output=all ::

12:53:30.71 [WARN] Failed to generate coverage data for python/projects/…/test.py:../integration-tests.

12:53:30.73 [WARN] Failed to generate JUnit XML data for python/projects/…/test.py:../integration-tests.

12:53:30.73 [ERROR] Completed: Run Pytest - python/projects/…/test.py:../integration-tests failed (exit code -15).
Exceeded timeout of 20.0 seconds when executing local process: Run Pytest for python/projects/…/test.py:../integration-tests:../integration-tests

✕ python/projects/…/test.py:integration-tests failed in 20.04s.

So, either using

--test-output=all

I cannot see logs 😞

pants version:

2.14.0a0

If I run the same command without test timeout,

./pants test --test-output=all ::

then I’ll see output of all tests. So, test timeout somehow remove the logs?

Hello. I have a problem with Pants in 2.13.0 when I try to use

.env

file and get the values with

python-dotenv

. Here is a quick example:

import os
from dotenv import load_dotenv

load_dotenv()

if __name__ == "__main__":
    print(os.environ["ENVVAR"])

I have a

.env

file with

ENVVAR=test

. When I

run

using Pants 2.12.0 it prints

"test"

without problem but I get this in Pants 2.13.0:

File "/tmp/pants-sandbox-9ZiWSR/tasks/helloworld/main.py", line 7, in <module>
    print(os.environ["ENVVAR"])
  File "/home/soheil/.pyenv/versions/3.10.6/lib/python3.10/os.py", line 679, in __getitem__
    raise KeyError(key) from None
KeyError: 'ENVVAR'

Any recommendation?

Yes. You'll see it every time pants encounters the

warn

in it's code. However work done is memoized in the daemon, so realistically you'll see it when you restart daemon or don't use it

IIRC Today, if a process is killed we don't capture the output. You might look an existing GitHub issue feature request or make your own

Thank you for the answer. It's not clear what that warning is printed, because I have files in that project, and other targets work fine. I'll see if I can figure it out 🙂

If you have a glob for a targets sources, and that glob fails to match any files that's when you should see the warning. It should be logging the targets' address

How are you expecting to see the file in the sandbox? Do you have an associated target?

Oh yeah actually I think I know what's going on. Previously the

run

sandbox was under the build root

.pants.d

but now it's under

/tmp

. You should declare a

file

target for your

.env

file and make sure the right targets depend on it, so it shows up in the sandbox

Yes, and no. I have a

.env

file in the root folder and a

tasks/

folder in which I have several Python project which are configured with environment variables. I deploy the PEX files in Amazon ECS in container images, and the environment variables is automatically set in the container in production. But during development I want to rely on an

.env

file.

Alright, I will try. Thanks!

We could also consider supporting .env as a plug-in. If you're feeling really ambitious

Hi! Is it possible to combine the default version of a package with a local one? It has been discussed before but I couldn't find a solution that suits my needs. It's about our friend

pytorch

. What we would like is to use the regular upstream version e.g.

1.12.1

for all OSes except Linux, where we want to use

1.12.1+cpu

. Our issue being that the upstream Linux wheel is very large and we don't use any of the capabilities included in the Linux wheel. The CPU one would suit us just fine. However, as soon as we add an index that contains any local versions it always only select a suitable package for the current platform (OS), ignoring all the other ones. This results in incompatible dependency error on other platforms than the one that generated the lockfile. The only solution that seems to work is to use the version upstream (pypi) but, as I said, then we unfortunately use the very large wheels for Linux. Any recommendations on how to solve / circumvent this issue?

You can also convert it to an error by setting this in pants.toml https://www.pantsbuild.org/docs/reference-global#unmatched_build_file_globs

[GLOBAL]
unmatched_build_file_globs = "error"

Hi, this was actually just improved this week! https://github.com/pantsbuild/pants/pull/16841 It looks like we need to cherry-pick it to 2.14 still. I will get that out today, and possibly another 2.14 rc

Woah cool!!

I had files in the places it warned about, that's why I was confused. I started migrating real projects and there's no warning. I will keep an eye on it, if I can reproduce it I will send more details. Thanks!