Back to the main page.

Bug 2285 - ft_appendfreq: no support for frequency structures with labelcmb

Reported 2013-09-13 16:28:00 +0200
Modified 2017-04-25 14:47:17 +0200
Product: FieldTrip
Component: core
Version: unspecified
Hardware: PC
Operating System: Mac OS
Importance: P3 normal
Assigned to: Robert Oostenveld
Depends on: 3287
See also:

Jan-Mathijs Schoffelen - 2013-09-13 16:28:40 +0200

it expects a label-field. one could probably work around this by temporarily creating a label field, but this is ugly and error prone (e.g. when channelcombinations are thrown out, the label field is updated, but the labelcmb field is not).

Jörn M. Horschig - 2014-04-16 10:52:28 +0200

so ft_checkdata(data, 'cmbrepresentation', 'full'), and after appending, 'sparse' is not good?

Jan-Mathijs Schoffelen - 2015-02-04 12:11:21 +0100

By now ft_appendfreq has been pimped. Nietzsche now runs into the issue, so I am reassigning it to her. The bottom line is that there is always a check to checkchan, which expects a label-field in the data. I would assume that falling back to checking the labelcmb (when no label is present) should work. TO DO: -think about whether the above assumption is correct -if so, make checkchan more robust

Robert Oostenveld - 2017-04-21 10:21:07 +0200

Arjen implemented ft_append_sens and made changes to ft_append_data. This made the inconsistency between appending data/timelock/freq very obvious. Please note that there is also an ft_appendsource, but I won't be looking at that one now.

Robert Oostenveld - 2017-04-25 14:47:17 +0200

The latest version of private/append_common supports covariance (i.e. chan_chan) but does not support chancmb. Do you have a test script, or could you come up with one?