[BUG] Drop annotations with NaN onset in EEGLAB raw files #12484
[BUG] Drop annotations with NaN onset in EEGLAB raw files #12484larsoner merged 10 commits intomne-tools:mainfrom
Conversation
|
@cbrnr Should be fixed now! Should I add that data to the testing dataset? |
|
I would not add a new dataset, but can you add a test with some artificially generated markers that demonstrate the problem (and verify the fix)? |
Would I have to make a temp file @cbrnr? Because from my understanding, I would have to read a MATLAB .set file then edit the annotations. Then I would need to read the file again to check to see if it drops the bad annotations. Can you think of another way? EDIT: I think I can just import EDIT2: I looked and EDIT3: Yeah, since MATLAB files are stored as a binary and the |
|
Hm, yes, because the fix is in the reader it's probably easiest if you add a small(!) test file then. |
Can you try this and see if it's easy? If it's old format then |
|
I can try @larsoner should I just use the |
|
No in tests typically you use the pytest test fixture |
|
That's a weird error -- I put a little fix in #12491 so ignore it for now and we'll merge that quickly and merge |
|
Just a couple of minor things, committed and marking for merge-when-green, thanks @withmywoessner ! |
|
Pip-pre is unrelated, merging! |
…12484) Co-authored-by: Eric Larson <larson.eric.d@gmail.com>

Reference issue
Fixes #12425.
What does this implement/fix?
Fix bug where MNE cannot handle EEGLAB events with NaN annotations
Additional information
Rationale for decision from EEGLAB documentation:

https://eeglab.org/tutorials/04_Import/Importing_Event_Epoch_Info.html