pynvml only semi optional

[maltevesper]

I would like to use pytest-isolate, however, as I have filterwarnings set to error in my pyproject.toml to make sure I do not ship warnings, pytest terminates with an Internal error

pyproject.toml pytest config:

filterwarnings = [
    "error",
]

Resulting output

INTERNALERROR> Traceback (most recent call last):
INTERNALERROR>   File "/workspaces/project/.venv/lib/python3.14/site-packages/pytest_isolate/plugin.py", line 62, in get_available_gpus
INTERNALERROR>     import pynvml
INTERNALERROR> ModuleNotFoundError: No module named 'pynvml'
INTERNALERROR> 
INTERNALERROR> During handling of the above exception, another exception occurred:
INTERNALERROR> 
INTERNALERROR> Traceback (most recent call last):
INTERNALERROR>   File "/workspaces/project/.venv/lib/python3.14/site-packages/_pytest/main.py", line 285, in wrap_session
INTERNALERROR>     config._do_configure()
INTERNALERROR>     ~~~~~~~~~~~~~~~~~~~~^^
INTERNALERROR>   File "/workspaces/project/.venv/lib/python3.14/site-packages/_pytest/config/__init__.py", line 1119, in _do_configure
INTERNALERROR>     self.hook.pytest_configure.call_historic(kwargs=dict(config=self))
INTERNALERROR>     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^
INTERNALERROR>   File "/workspaces/project/.venv/lib/python3.14/site-packages/pluggy/_hooks.py", line 534, in call_historic
INTERNALERROR>     res = self._hookexec(self.name, self._hookimpls.copy(), kwargs, False)
INTERNALERROR>   File "/workspaces/project/.venv/lib/python3.14/site-packages/pluggy/_manager.py", line 120, in _hookexec
INTERNALERROR>     return self._inner_hookexec(hook_name, methods, kwargs, firstresult)
INTERNALERROR>            ~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
INTERNALERROR>   File "/workspaces/project/.venv/lib/python3.14/site-packages/pluggy/_callers.py", line 167, in _multicall
INTERNALERROR>     raise exception
INTERNALERROR>   File "/workspaces/project/.venv/lib/python3.14/site-packages/pluggy/_callers.py", line 121, in _multicall
INTERNALERROR>     res = hook_impl.function(*args)
INTERNALERROR>   File "/workspaces/project/.venv/lib/python3.14/site-packages/pytest_isolate/plugin.py", line 336, in pytest_configure
INTERNALERROR>     register_resource_provider("gpu", "CUDA_VISIBLE_DEVICES", get_available_gpus)
INTERNALERROR>     ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
INTERNALERROR>   File "/workspaces/project/.venv/lib/python3.14/site-packages/pytest_isolate/resource_management.py", line 153, in register_resource_provider
INTERNALERROR>     existing = provider_func()
INTERNALERROR>   File "/workspaces/project/.venv/lib/python3.14/site-packages/pytest_isolate/plugin.py", line 70, in get_available_gpus
INTERNALERROR>     warnings.warn("Failed to get GPU count using pynvml", RuntimeWarning)
INTERNALERROR>     ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
INTERNALERROR> RuntimeWarning: Failed to get GPU count using pynvml

It would be nice to have a configuration option to tell the plugin that one does not care about the GPU parts.

[gilfree]

Would you be willing to add such configuration?

I am not sure if it is doable in a simple manner - not sure if the configuration is available early enough.

I am also ok with dropping the warning at all, or having a custom warning type.

If you have other suggestions - I'm open.

[maltevesper]

There is a workaround, but this feels clunky, adding a warning ignore: "ignore:Failed to get GPU count using pynvml:RuntimeWarning:pytest_isolate", to the pyproject.toml.

[maltevesper]

Would you be willing to add such configuration?

I am not sure if it is doable in a simple manner - not sure if the configuration is available early enough.

I am also ok with dropping the warning at all, or having a custom warning type.

If you have other suggestions - I'm open.

Wow, this was fast. seeing that the code owner is this responsive definitly raises the priority. I have to do the fix in my spare time though, I will see what I can do.

[gilfree]

You just got lucky 😄 that I was doing something related just now.