WIP FIX Deprecate use_rounding=False default in time_as_index [circle full] – take 2

[mathias-sm]

Reference issue (if any)

Implements the changes suggested in #13634 to estimate the impact of defaulting to truncation to defaulting to rounding when converting time to indices.

This is a copy / continuation of #13848 which I accidentally closed by moving my changes to branch fix-issue-13634-rounding and reverting my main to the upstream's main so that I could work on another PR. It'll teach me a lesson, sorry for the noise.

What does this implement/fix?

time_as_index defaulted to truncation (use_rounding=False), causing get_data(tmin=...) to return different samples from crop(tmin=...).get_data() at certain times due to floating-point truncation. Change the default to None and rounding. Emits a FutureWarning prompting callers to pass True or False explicitly. The goal of this commit is to estiate the impact of this change; follow-up commit should internally update all use of time_as_index to explicitely set a desired default.

Additional information

I have not yet updated all of the internal use of time_as_index, some of which explicitely use either use_rounding=False or use_rounding=False. Internall calls that do not pass a value for use_rounding will emit warnings. This is not ready for merging and is submitted as WIP / as a draft PR to trigger all tests and estimate the impact of this change.

What do we know so far?

From the previous discussion, we know that:

• In the existing tests, mne/io/egi/tests/test_egi.py::test_egi_mff_pause is the only test that breaks with this change: it boils down to onset/offsets comparisons that were hardcoded and are therefore now off-by-one.
• Temporarily, the Warnings are suppressed because those break tests. If we go ahead with these changes, I will have to either update the tests to catch those warnings or to set explicit use_rounding to avoid them in the first place.
• There are about 29 places in the codebase where time_as_index is called without explicitely setting use_rounding, which would need updating if we go ahead with this change.
• As far as I can tell nothing else was broken.

[welcome]

Hello! 👋 Thanks for opening your first pull request here! ❤️ We will try to get back to you soon. 🚴

[larsoner]

To really see the impact I would suggest following my RuntimeError suggestion from #13634 (comment)

By computing both ways and raising this error it would clearly show all tests and examples (with a [circle full]) there the behavior would actually change.

[mathias-sm]

Ah yes indeed somehow I had forgotten about this. Quick exploration of that: it breaks 5 tests. I will push this after making sure that I'm doing things right to get them from the CI.

FAILED mne/io/tests/test_raw.py::test_get_data_tmin_tmax - RuntimeError: This would have returned a different value: index=array([600.61499023]), different from np.round(index)=array([601.]).
FAILED mne/stats/tests/test_erp.py::test_compute_sme - RuntimeError: This would have returned a different value: index=array([180.06149902]), different from np.round(index)=array([180.]).
FAILED mne/tests/test_evoked.py::test_get_peak - RuntimeError: This would have returned a different value: index=array([30.03074951]), different from np.round(index)=array([30.]).
FAILED mne/viz/tests/test_ica.py::test_plot_ica_overlay - RuntimeError: This would have returned a different value: index=array([1801.8449707]), different from np.round(index)=array([1802.]).
FAILED mne/tests/test_proj.py::test_compute_proj_raw - RuntimeError: This would have returned a different value: index=array([1501.53747559]), different from np.round(index)=array([1502.]).

[larsoner]

... and CIs won't run until the style runs are happy!

Thanks for working on this

[larsoner]

Looks like there are a number of failures:

= 56 failed, 6180 passed, 29 skipped, 109 deselected, 8 xfailed, 2 xpassed in 3609.83s (1:00:09) =

but a whole bunch are a single parametrization of mne/preprocessing/tests/test_realign.py, so it's really more like ~12 failures / tests that need to be updated. Want to try to fix those next?