Hey all, I'm deploying a Django application via Pants to AWS Lambda, and I'm seeing this error in our Cloudwatch logs:

[ERROR] TypeError: the 'package' argument is required to perform a relative import for '.app.lambdex_handler'
Traceback (most recent call last):
  File "/usr/local/lib/python3.9/importlib/__init__.py", line 122, in import_module
    raise TypeError(msg.format(name))

Has anyone seen this before? There must be some kind of Python versioning issue, but I'm not sure what's going on. Our Lambda runtime is running Python 3.9.

📣 New! Meet the Pants team members, and discover what each of us is proudest of contributing to the project. For in-depth introductions, see also the team member self-introduction series we've recently started in #welcome as well as the "Meet our newest Maintainer" series over on the blog, including this brand new profile just posted about @witty-family-13337: https://blog.pantsbuild.org/meet-our-newest-maintainer-a-alonso-dominguez/

Can Pants be configured to emit a warning whenever it merges the path of two modules/packages? Especially if it's first vs. thirdparty?

Is there a simple command to check if each file has only one owner?

I am looking at this code and trying to understand if the SHA256 mentioned on lines from 13 to 16 are the ones for binaries hosted on URL: https://binaries.pantsbuild.org/bin/grpc_python_plugin/{version}/{platform}/grpc_python_plugin? Should these SHA256 match for a given platform if I built grpc version 1.32.0 from its source and then check shasum of grpc_python_plugin file?

Hi I am trying to debug some custom pants plugin using Process. It just returned “no such file or directory”. How could I know what file or directory it is looking for?

I am running the fmt goal and observing a discrepancy that fmt is not picking up the target when run on all targets, but works when a precise file is given. %

./pants

fmt

::

... ✓ black made no changes. ✓ docformatter made no changes. ✓ isort made no changes. <--- HERE %

./pants

fmt

projects/appsflyer/appsflyer/apps/appsflyer/views.py

... ✓ black made no changes. ✓ docformatter made no changes. + isort made changes. <--- HERE %

./pants

list

::

... projects/appsflyer/appsflyer/apps/appsflyer/views.py <-- THE TARGET IS PRESENT ...

I’m developing a microservice using tensorflow and pytorch with mono-repo. Building this directory consumes roughly 100GB of disk. I checked the details and it seems that most of it is consumed as cache. I would like to limit the cache as the disk for github actions specification is limited. https://docs.github.com/en/actions/using-github-hosted-runners/about-github-hosted-runners#supported-runners-and-hardware-resources is it possible to configure such a setting?

Hey, is there a way to mark all python modules used in pants as typed ? eg. by enforcing to add a

py.typed

file or adding it implicitly for mypy and packaging?

Pex build is failing, please help if you know the cause and how to deal with it. I have specified pex_binary.complete_platforms.

$ ./pants package path/to/pex:bin
...
1. cp38-cp38-manylinux_2_27_x86_64:
    Failed to resolve all requirements for complete platform cp38-cp38-manylinux_2_27_x86_64 from thirdparty/python/lockfiles/python_default.txt:

Configured with:
    build: False
    use_wheel: True

Dependency on promise not satisfied, 1 incompatible candidate found:
1.) promise 2.3 (via: tensorflow-datasets<5.0.0,>=4.6.0 -> promise) does not have any compatible artifacts:
    <https://files.pythonhosted.org/packages/cf/9c/fb5d48abfe5d791cd496e4242ebcf87a4bb2e0c3dcd6e0ae68c11426a528/promise-2.3.tar.gz>

Has anyone used gunicorn + pex? I had a search in the channel and looks like a few have... I'm having issues getting the pex to bundle up my actual app with error

ModuleNotFoundError: No module named 'main'

Are there any ways to debug this? Open up a Pex??

Another reason my file-based deps proposal makes the world more sane. A Python tool depending on a shell script depending on a PEX Binary is erroring about multiple resolves since my py tool uses resolve A and my pex uses resolve B. We're inheriting the PEXs deps, even though we're using a built artifacts 💀 😭

Hi, is it possible to lock a specific transitive dependency ? When running

./pants test ::

I am running into an issue with protocol buffers that require me to lock them to 3.20.x, I already do in one requirements file at the root level. packages inside src/ pick the dependency, but tests don’t

Weird PEX issue involving

awscli

(happens in any

layout

mode). Pex attached...

$ python3.8 aws.pex 
Traceback (most recent call last):
  File "/home/joshuacannon/.pex/venvs/55691004f59e3a2b4d0b5c07b229db01742aa7db/779eb2cc0ca9e2fdd204774cbc41848e4e7c5055/bin/aws", line 21, in <module>
    import awscli.clidriver
ModuleNotFoundError: No module named 'awscli'

Hi guys. We are trying to get Django running with pants. Is there a way to have pants scan all python files in a folder to determine dependencies instead of traversing the tree?

Hi everyone. I'm trying to understand the following error while trying to run a Django app via AWS Lambda:

[ERROR] ResolveError: Failed to resolve requirements from PEX environment @ /app.
Needed cp38-cp38-manylinux_2_26_x86_64 compatible dependencies for:
 1: pillow>=6.2.0
    Required by:
      matplotlib 3.5.1
    But this pex had no ProjectName(raw='pillow', normalized='pillow') distributions.
 2: pillow~=9.2.0
    raise ResolveError(/pex/environment.py", line 589, in resolve_distsp_pex_envions.

I have

pillow ~= 9.2.0

and

matplotlib==3.5.1

as dependencies in my requirements.txt file. When I build my Lambda pacakge locally, I see

Pillow-9.2.0-cp38-cp38-manylinux_2_28_x86_64.whl

in my .deps directory. I figured Pillow ~= 9.2.0 would match matplotlib's Pillow version requirements? What am I doing wrong?

Is there a way within

pants.toml

to specify a default for an env var if it's not set in the caller's environment?

I'm trying to add a test to my package in order to check if specific python (sub-)packages are namespace packages. For the top level package, it would look like this:

mypkg/
  subpkg1/
    __init__.py
  subpkg2/
    __init__.py

Notice the missing

__init__.py

under

mypkg

, which makes it an implicit namespace package (PEP-420). Now the test I wrote looks like this:

# tests/test_namespace.py
import mypkg

def test_namespace():
    assert mypkg.__file__ is None

While this test runs successfully in pytest, it fails in the hermetic environment of pants, because the dependency on

mypkg

does not include any files. The directory

mypkg

itself is not created in the local environment, so

mypkg

is not importable. I know there are a couple of trivial solutions to this (

import mypkg.subpkg1

in the test, specifying a dependency manually or checking for the absence of

mypkg/__init__.py

in pre-commit hooks), but is there any solution in pants that I'm missing, e.g. some configuration option?

We're actively growing the Pants team all the time. We've welcomed 5 new Maintainers and 12 new Contributors in the past year alone! One way you can help the project continue to flourish is by pointing out folks in our community who show potential for becoming good future team members. The entry level criteria are a demonstrated commitment to Pants, and some track record of contributions to the project — manifested in any of many ways, not necessarily only via code. When someone does standout work, or shows special enthusiasm for giving back to the project, we don't want to let that fall through the cracks. We want to work with them to help meet mutual goals. So feel free to DM me anytime to point out someone to have on the radar. Thanks so much!

Is there a way to make pants use a pager via an env var? I’m trying to record an asciinema of

./pants help goals

but it scrolls by too quickly.

Hey all! I was wondering how to accomplish something like bazel's new_local_repository in pants -- in particular: • I have some system libraries that were built/installed without pants • I want to tell pants that they exist, so that a pants target can depend on them How would I do this in pants? Or is there an alternative way to accomplish the same thing?

Is it on purpose that pex_binary can't include shell_sources or am I messing up somewhere? I have a tool that sub processes out to more shell code than Python and want to use the pex machinery to package it into a nice standalone bin. Do I have to make them resources? I remember there was discussion to remove the restriction on non-rc files because facilities like Path(file) make it easy to avoid the pitfall noted in the docs—was a design consensus reached on that? I also didn't know other target types would fall under the files target umbrella—I take it you only have python sources and resources?

Hi all! I’d like to gather some information about Docker use at build time: A) How, if at all, do you currently have Pants and Docker interacting, and B) What would you like that interaction to look like in an ideal world? [Note that I’m not talking about using Pants to build Docker images, but on using Docker at build time for consistency]. Please reply to the thread or DM me if you prefer. Really appreciate your input, which will go towards Pants improvements in this area.

Any having issues with VSC git requirements.txt with basic auth randomly failing to authenticate when packaging?

%

./pants fmt ::

192207.29 [INFO] Completed: Format with docformatter - docformatter made no changes. 192207.29 [WARN] Completed: Format with isort - isort made changes. /path/to/file1 /path/to/file2 192207.29 [WARN] Completed: Format with Black - black made changes. /path/to/file1 /path/to/file2 + black made changes. ✓ docformatter made no changes. + isort made changes. But I can't see any files changing in the git in vscode?

Note that the Pants blog is temporarily down as we upgrade to Ghost 5.0

I have a pyproject.toml that depends on some local wheels (internal 3rdparty deps). Whenever I generate the lockfile (pex) I see that it uses the paths on my local computer… this means that I can’t really commit that lockfile to git (and thus have to manually regenerate on CI). Is there a way to get around this? Would the answer be to use a different resolve for local wheels and only generate that lockfile every time?

I'd like to set up a

python_distribution

with all the files (recursively) in a directory as contents. Is there a more maintainable way to do that than just injecting the contents of

pants list dir::

into the

dependencies

field?

Hi, is it possible to install 1st party python packages in editable mode using pants? I know there isn’t a specific goal for this but I was wondering what would be the way to do it when using pants, if possible.

Hi, has anyone combined

nix

and

pants

? I’d be interested in that, what did you do and how did it work out for you?