DeprecationWarning regarding the use of LegacyVersion

[jtilahun]

Describe the bug
The below DeprecationWarning appears since installing pandas-stubs:

07:59:41.56 [WARN] /home/jtilahun/.cache/pants/setup/bootstrap-Linux-x86_64/2.15.0_py38/lib/python3.8/site-packages/pkg_resources/_vendor/packaging/specifiers.py:255: DeprecationWarning: Creating a LegacyVersion has been deprecated and will be removed in the next major release
  warnings.warn(

To Reproduce

1. Install pandas-stubs via pip by running pip install pandas-stubs from within the virtual environment.
2. Run a Python test. Using the Pants build system, this can be done as shown below:

$ ./pants test ops/utils/rpc_utils_test.py
07:59:41.56 [WARN] /home/jtilahun/.cache/pants/setup/bootstrap-Linux-x86_64/2.15.0_py38/lib/python3.8/site-packages/pkg_resources/_vendor/packaging/specifiers.py:255: DeprecationWarning: Creating a LegacyVersion has been deprecated and will be removed in the next major release
  warnings.warn(

07:59:47.18 [INFO] Completed: Run Pytest - ops/utils/rpc_utils_test.py:tests - succeeded.
Partition: ops_unit_tests

✓ ops/utils/rpc_utils_test.py:tests succeeded in 5.38s.

Wrote test reports to dist/test_results

Please complete the following information:

• OS: Linux
• OS Version Ubuntu 20.04
• python version 3.8.10
• version of type checker pyright 1.1.295
• version of installed pandas-stubs pandas-stubs 1.5.3.230321

Additional context
I never saw this warning prior to installing pandas-stubs, but this warning shows up constantly now. It's difficult to trace the contents of the warning message directly to pandas-stubs though.

[Dr-Irv]

We'd need a simple test case in order to replicate.

Looking at where the warning comes from, it might be due to how we specify the dependencies of pandas-stubs and how that specification interacts with pants. But in order to investigate, we'd have to be able to replicate this.

[jtilahun]

The simplest test case is in fact a stub unit test as shown below:

import unittest


class Test(unittest.TestCase):
    def test(self):
        self.assertTrue(True)

with the stub test saved as test.py and the test target declared in BUILD.pants as:

python_test(
    name="test",
    source="test.py",
)

Running ./pants test on this Python test suffices to make the warning appear:

$ ./pants test ops/utils:test
10:24:58.81 [WARN] /home/jtilahun/.cache/pants/setup/bootstrap-Linux-x86_64/2.15.0_py38/lib/python3.8/site-packages/pkg_resources/_vendor/packaging/specifiers.py:255: DeprecationWarning: Creating a LegacyVersion has been deprecated and will be removed in the next major release
  warnings.warn(

10:24:58.85 [INFO] Completed: Run Pytest - ops/utils:test - succeeded.
Partition: ops_unit_tests

✓ ops/utils:test succeeded in 1.58s (memoized).

Wrote test reports to dist/test_results

I think the warning is emitted during the stage where Pants is doing its resolution, because it looks like the warning appears before Pytest is even invoked.

[twoertwein]

You could probably run python -Werror .... to get a trace of where the warning is coming from.

[jtilahun]

Hmm, there was actually a typo in requirements.txt. Changing

pandas-stubs>=1.5.3<2.0.0

to

pandas-stubs>=1.5.3,<2.0.0

in requirements.txt and regenerating constraints appears to make the warning go away. Having said this, I'm unsure of how a typo in requirements.txt leads to a DeprecationWarning of all things. The constraints generation did not complain when the typo was present, which confuses me. Is there an actual deprecation present somewhere or is this a misleading warning? If there is an actual deprecation, the deprecation should still be addressed. However, if there are actually no deprecations, I guess that this may be simply be a case of a warning being misleading.

[Dr-Irv]

Based on this issue in pants pantsbuild/pants#17848, I think this is what happened. Nothing to do with pandas-stubs. Closing