instability in 2023.1 when loading studies
alenarto opened this issue · comments
Description
We're having issues of instability with eeglab 2023.1 that are difficult to pinpoint. Yesterday we ran into issues loading a study, with errors never seen before such as about having issues with function pop_loadset.m, e.g., a Matlab internal error saying that there's a function lacking the "end" statement.
I couldn't localize the error (as we've never modified these functions and they worked before), so I completely re-installed eeglab2023.1.
It worked for a moment (the previously unloadable study was now loaded), and then upon a future loading of the study it started spitting out errors such as in the error link. Error 2 ("Reference to non-existent field 'novodatchans'" in insertchans() at line 321) was reproduced by two people working on two different studies.
#### Steps to Reproduce
- I am sorry but this is difficult to reproduce. For us it seems to require loading the study, any study. I've added a small study into the drive at link above as a sample of a study that generated error2 on the last attempt (but it also did generators the other errors, but didn't generate them upon reinstall of 2023.1).
#### Expected behavior: A study to load.
#### Actual behavior: Variable unsystematic errors are produced as above. Seems to happen after several users use the same version of this eeglab. Ok for a little while after a reinstall.
#### Versions
OS version | [Centos7 on HCP cluster] |
Matlab version | [9.8.0.1396136 (R2020a) Update 3] |
EEGLAB version | [2023.1] |
I believe the first 2 errors are due to a typo (inserted by you :-) in eeg_checkchanlocs.m because I have checked the code and there is no "novodatchans" variable in the code.
For the last error, I have added another check to the code. Let me know.
Hi Arno. For "novodatchans" you're right - the corrupt version contains a typo that someone inadvertently inserted. The other errors are a little trickier but we've reinstalled and will report if we encountered them again. THanks!
OK, thanks. Let me know