Issue with icadefs.m being in ERPLAB

[jnvandermeer]

If you call binica from the gui,

then icadefs is called, which sets ICABINARY:

ICABINDARY = fullfile(eegpab_p, 'functions','supportfiles','ica_linux')

but then, binica.m is called (which then calls the ICABINARY), but there, it is written:

line 115: icadefs (this sets ICABINARY to the full path of where-to-find ica_linux executable). line 117: ICABINDARY = fullfile(eegpab_p, 'functions','supportfiles',ICABINARY) (this messes up the full path of where to find ica_linux, by adding yet again the full path of eeglab, etc)

... which effectively makes ICABINARY unfindable.

If I change binica.m, on line 117, and replace ICABINARY with 'ica_linux', it works again.

I think this is a bug in eeglab, and may only be with linux, and not (very) often noticed because everybody just uses runica.m

[arnodelorme]

I think this issue has been fixed because icadefs.m no longer has the code setting the full path

ICABINDARY = fullfile(eegpab_p, 'functions','supportfiles','ica_linux')

[jnvandermeer]

Hi Arnoud, you are right!

It was ERPLAB10.04, the eeglab plugin, that also contained and icadefs.m file. Because I added this plugin, the icadefs of ERPLAB was used, which lives in plugins/ERPLAB10.04/studio_functions/Functions/EEGLAB/Scalp/icadefs.m.

So, this probably better belonged to the ERPLAB repository.

[arnodelorme]

Interesting point, one more reason to retire icadefs.m (which I was planning to do). I will implement a fix.

[arnodelorme]

@guanghuizhang0328 and @davidg622, how can we avoid the problem above?

[arnodelorme]

This was fixed in 1347f9842