Back to the main page.

Bug 1433 - test_ft_datatype_sens fails

Reported 2012-04-18 15:15:00 +0200
Modified 2012-12-31 11:46:25 +0100
Product: FieldTrip
Component: core
Version: unspecified
Hardware: PC
Operating System: Windows
Importance: P3 normal
Assigned to: Robert Oostenveld
Depends on:
See also:

Johanna - 2012-04-18 15:15:06 +0200

It first fails for i=3 fullfile('neuromag122', 'jg_single_01raw.fif'). It seems from SVN revision 4670, there was a check in test_ft_datatype_sens that if 'unit' not exist on sens2, then add the 'unit' of sens1. This works for all data cases except i=3, because this data is empty (all the fields exist, but are size 0x3). ft_convert_units requires that fields not be empty. Perhaps a bigger question is, why do both sens1 and sens2 for i=3 have empty fields? It secondly fails for i=16 fullfile('yokogawa440', 'S1_MEG_Epoch.raw'). sens1 (on disk) shows normal values, but sens2 some fields are zero (e.g. chanori is the correct size but all zeros). I traced it to line 99 of ft_datatype_sens where sens.ori makes sense but then the output of chanori is all zeros. I think Robert may know the answer to i=3, and JM for i=16 ...?

Johanna - 2012-04-18 15:17:17 +0200

*** Bug 1417 has been marked as a duplicate of this bug. ***

Robert Oostenveld - 2012-12-31 11:25:32 +0100

The test does not show up as failed on the dashboard, since it has a try-catch and then shows a warning instead of an error. I reran the test and it fails for K>> i i = 16 K>> filename(i) ans = 'yokogawa440/S1_MEG_Epoch.raw' The failure is due to chanpos and ori being all zeros for sens2

Robert Oostenveld - 2012-12-31 11:39:30 +0100

(In reply to comment #2) correction: only the chanori is all zeros in sens2, but the chanpos is still different between sens1 and sens2.

Robert Oostenveld - 2012-12-31 11:44:15 +0100

There only exists one 440-channel yokogawa system in the world and it has a complicated mixture of channel types. I would rather not spend time on debugging this system at the moment. I have removed the warning and made it into an error, that makes the script show up in the dashboard. I have also updated the test so that the known discrepancy for yokogawa440 does not trigger an error. roboos@mentat001> svn commit test/test_ft_datatype_sens.m Sending test/test_ft_datatype_sens.m Transmitting file data . Committed revision 7273.

Robert Oostenveld - 2012-12-31 11:46:25 +0100

closed several bugs that have been resolved for some time. Feel free to reopen the bug if you disagree.