BUG: warn magic isn't magical enough for contextlib.contextmanager

[larsoner]

... I was wondering why the log contains next(self.gen):

>>> raw = mne.io.read_raw("/Users/clemens/Downloads/S01_MI/motorimagination_subject1_run1.gdf")
Extracting EDF parameters from /Users/clemens/Downloads/S01_MI/motorimagination_subject1_run1.gdf...
GDF file detected
Setting channel info structure...
Could not determine channel type of the following channels, they will be set as EEG:
F3, F1, Fz, F2, F4, FFC5h, FFC3h, FFC1h, FFC2h, FFC4h, FFC6h, FC5, FC3, FC1, FCz, FC2, FC4, FC6, FTT7h, FCC5h, FCC3h, FCC1h, FCC2h, FCC4h, FCC6h, FTT8h, C5, C3, C1, Cz, C2, C4, C6, TTP7h, CCP5h, CCP3h, CCP1h, CCP2h, CCP4h, CCP6h, TTP8h, CP5, CP3, CP1, CPz, CP2, CP4, CP6, CPP5h, CPP3h, CPP1h, CPP2h, CPP4h, CPP6h, P3, P1, Pz, P2, P4, PPO1h, PPO2h, eog-l, eog-m, eog-r, thumb_near, thumb_far, thumb_index, index_near, index_far, index_middle, middle_near, middle_far, middle_ring, ring_near, ring_far, ring_little, litte_near, litte_far, thumb_palm, wrist_bend, roll, pitch, gesture, armeodummy, armeodummy, armeodummy, armeodummy, armeodummy, armeodummy, armeodummy, armeodummy, armeodummy, armeodummy, armeodummy, armeodummy, armeodummy
Creating raw.info structure...
/Users/clemens/.local/share/uv/python/cpython-3.12.7-macos-aarch64-none/lib/python3.12/contextlib.py:144: RuntimeWarning: Channel names are not unique, found duplicates for: {'armeodummy'}. Applying running numbers for duplicates.
  next(self.gen)

Originally posted by @cbrnr in #12909 (review)

[larsoner]

Probably not just your setup -- our warn does some magic to set the stacklevel such that it's out of the mne namespace. My guess is that some code is using a @contextlib.contextmanager and while traversing the stack, contextlib is outside the mne namespace so our warn stops there and emits a warning. Instead it should know contextlib is special in this way and keep going.

[larsoner]

I can't replicate on my system:

import contextlib
import mne


@contextlib.contextmanager
def my_warn():
    yield


with my_warn():
    mne.io.read_raw_fif(mne.datasets.testing.data_path() / "SSS" / "test_move_anon_raw.fif", allow_maxshield=True)

gives a correct line 11, not anything in contextlib:

/home/larsoner/Desktop/warncheck.py:11: RuntimeWarning: This file contains raw Internal Active Shielding data. It may be distorted. Elekta recommends it be run through MaxFilter to produce reliable results. Consider closing the file and running MaxFilter on the data.
  mne.io.read_raw_fif(mne.datasets.testing.data_path() / "SSS" / "test_move_anon_raw.fif", allow_maxshield=True)