Back to the main page.

Bug 1707 - channel order shouldn't matter for the output of an ft_function

Status CLOSED FIXED
Reported 2012-09-12 14:16:00 +0200
Modified 2016-05-09 08:49:37 +0200
Product: FieldTrip
Component: core
Version: unspecified
Hardware: PC
Operating System: Windows
Importance: P3 normal
Assigned to: Jan-Mathijs Schoffelen
URL:
Tags:
Depends on:
Blocks:
See also:

Jan-Mathijs Schoffelen - 2012-09-12 14:16:13 +0200

Anecdotally, Stephen reported issues relating to different channels and different orderings of channels that determine the output of ft_timelockstatistics, depending on what the order of the input arguments has been. FieldTrip should nowhere assume that the order and list of channels present is the same for all data input arguments, nor that the order is the same in e.g. the data.label and data.grad.label. TODO; make a test script that inputs stuff and outputs stuff


Jörn M. Horschig - 2012-09-20 14:43:50 +0200

started a test-script - loads of work


Jan-Mathijs Schoffelen - 2012-09-26 14:11:05 +0200

This goes for all functions that take multiple data structures in the input.


Robert Oostenveld - 2013-10-28 14:30:31 +0100

I have commented out a part of the existing test code, which was clearly incomplete. This should at least allow this script not to show up as non-informative error on the dashboard. mac001> svn commit test_bug1707.m Sending test_bug1707.m Transmitting file data . Committed revision 8660.


Robert Oostenveld - 2014-02-21 15:45:50 +0100

mac001> svn commit Deleting test_bug1707.m Committed revision 9218. I removed an incomplete test script. Please see http://code.google.com/p/fieldtrip/source/browse/trunk/test/test_bug1707.m?spec=svn7157&r=7157 if you want to retrieve it.


Jörn M. Horschig - 2014-12-22 14:27:05 +0100

being realistic, I will never finish this bug as it is way too much work. Also, FT seems to be stable wrt channel ordering now. Assigning back to default in case someone else wants to do this.


Jan-Mathijs Schoffelen - 2016-02-22 09:17:45 +0100

I think that over the past year, quite some things have been ironed out. For a start, the more or less consistent use of ft_selectdata results in more strict bookkeeping and channel order checking. Since this bug is predating the ft_selectdata era, and because it is rather 'generic' I suggest to close for now, and take up any channel order related issues once they concretely arise.