Back to the main page.

Bug 989 - there seems to be a problem in selectdata in that the underlying dimlength does not recognize a .stat field.

Reported 2011-09-22 19:24:00 +0200
Modified 2011-10-05 15:08:12 +0200
Product: FieldTrip
Component: core
Version: unspecified
Hardware: PC
Operating System: Mac OS
Importance: P1 normal
Assigned to: Jan-Mathijs Schoffelen
Depends on:
See also:

Jan-Mathijs Schoffelen - 2011-09-22 19:24:22 +0200

1) dimlength only recognizes hardcoded fields, (cov, powspctrm), not user specified, e.g. 'stat' which would be common in a 'second level statistic'. So in the case I enter a series (becoming the rpt) of datastructures in freqstatistics, all containing dimord = 'chan_freq', with only a .stat field as the data, it cannot infer the dimensionality (i.e. dimlength). However, freqstatistics was called with cfg.parameter = 'stat'. This is however not passed over to dimlength in selectdata. Problem occurs (recursively) at dimlength line 46 and 116 2) Also, it there are problems swallowing one-dimensional data. Still is high on my wishlist since now i have to add a dimension (_freq) and an extra field (.freq). Please let me know if I need to elaborate. Cheers, Stephen

Jan-Mathijs Schoffelen - 2011-09-22 19:25:15 +0200

Hoi Stephen, Ik heb een aparte bug hiervoor aangemaakt, en assign deze (=989) weer aan jou ;o), want dit betreft het test script

Jan-Mathijs Schoffelen - 2011-09-22 20:00:56 +0200

This is not true, it works when you call ft_selectdata(data{:}, 'param', 'stat')

Stephen Whitmarsh - 2011-09-22 20:09:25 +0200

Nietes! Morgen, om 9 uur bij het fietsenrek!

Jan-Mathijs Schoffelen - 2011-09-22 20:19:06 +0200

This is a difficult one; if dimlength has to estimate the length 'rpt', it does not have knowledge in which field to look, so it seems the most prevalent fields should be tested. I added .stat for the time being

Stephen Whitmarsh - 2011-09-22 20:23:04 +0200

Thanks, much obliged! I knew we could work it out :)