Back to the main page.

Bug 1200 - matlab crashing when starting to use FieldTrip

Reported 2011-11-30 11:24:00 +0100
Modified 2012-06-20 15:03:20 +0200
Product: FieldTrip
Component: core
Version: unspecified
Hardware: Other
Operating System: Linux
Importance: P3 critical
Assigned to: Boris Reuderink
Depends on:
See also:

Lilla Magyari - 2011-11-30 11:24:48 +0100

Hi, I had already the same problem (see bug 1068) a month ago, and now the same thing happens again. (It is probable that I haven't checked last time, if the problem disappeared because I have already started to use an older version of FieldTrip due to the problem. So, it is just possible that the problem hasn't been solved.) When I try to start to use FieldTrip (28.11.2011 version) it is crashing. I attached the script that I used. Addpath, ft_defaults works fine, but when I try to run ft_sourcegrandaverage I get a system error and matlab crashes (but last time this happened with ft_read_mri function, so I suspect it doesn't depend on the actual function used). I am using Matlab2008b on the MPI grid. I do not have this problem with the earlier version of FieldTrip (like 2011 01 26). Lilla

Lilla Magyari - 2011-11-30 11:51:08 +0100

I tried the script on a 2010b version of Matlab, and it is running! But I could try this because 2010b is installed locally on my PC. At the MPI, we can use only Matlab 2008b on the grid. (I have asked, they haven't installed any newer version). Lilla

Boris Reuderink - 2011-12-06 13:00:43 +0100

Dear Lilla, this sounds quite severe. But I can't find an attachment with the script. Could you add it once again? And could you describe the context for the problem in a bit more detail (specifically, which operating system do you use)?

Lilla Magyari - 2011-12-07 11:09:48 +0100

(In reply to comment #2) Hi Boris, I have tried to confirm the error also on the Donders mentat with the same version of Matlab that we use at the MPI (2008Rb). But it works just fine. So, it is not because of version of Matlab we use here. I wrote an email to Herbert Baumann who is responsible for the grid, and matlab installation at the MPI...etc. about it. Maybe we can wait until he reacts on it. Lilla

Boris Reuderink - 2012-01-03 12:19:01 +0100

Hi Lilla, Any news on this issue? Best, Boris

Lilla Magyari - 2012-01-09 15:19:35 +0100

(In reply to comment #4) hey, our system administrator has upgraded the linux system, but I can't check yet if FieldTrip works properly now, because I can't login to our linux machines due to some after-effect of the upgrade :). Hopefully, I can do it in a few days. I'll let you know. Best, Lilla

Lilla Magyari - 2012-01-13 12:01:01 +0100

(In reply to comment #4) Hi Boris, our linux system has been updated (it is linux 64 using enterprise version 10, service pack 4) but I still have the same problem. I am clueless why.

Boris Reuderink - 2012-01-13 13:14:18 +0100

Hmm, I don't know either. Could you send me your script, and the output you can get before MATLAB crashes?

Lilla Magyari - 2012-01-26 14:42:25 +0100

Created attachment 218 output of crash

Lilla Magyari - 2012-01-26 15:05:53 +0100

(In reply to comment #8) hey, you can find data and script in /home/language/lilmag/beamforming_stat The script you find there crashes under my matlab at the mpi grid. We use matlab 2008b, but when I try the exactly the same script on the donders mentats with matlab 2008b, it is running perfectly. You can find attached the output of the crash. And this is only one example because the crash happens basically any time, the matlab encounters a fieldtrip function. (Except of ft_defaults. I can get only until here.) I tried also that I did not do addpath.... and ft_defaults (either in matlab command window or in startup.m) but I set the path myself in the menu with "add folder with its subdirectories". But I got the same error. Interestingly, older fieldtrip versions (for example, 2011 06 23) are running fine. I do not know which fieldtrip version is the earliest that I can not run anymore. It is not only at the mpi who has this problem, so it doesn't depend on my personal settings of matlab. (but may there is some default setting here why it is not working.) Lilla

Lilla Magyari - 2012-02-01 14:04:27 +0100

Created attachment 219 latest output of crash

Lilla Magyari - 2012-02-01 14:05:47 +0100

Created attachment 220 script used to produce the crash

Lilla Magyari - 2012-02-01 14:06:36 +0100

Created attachment 221 data used to produce the crash

Lilla Magyari - 2012-02-01 14:07:41 +0100

in the test script the first part (ft_convert_units) run OK, but the ft_read_headshape function caused a crash

Boris Reuderink - 2012-02-01 14:27:51 +0100

Lilla demonstrated the problem, as such it is confirmed. However, reproducing this bug is more difficult; the adapted test script does run on mentat301.

Boris Reuderink - 2012-02-01 15:57:04 +0100

Robert deduced that it probably does not involve mex files, and that the problem relates to this specific MATLAB and operating system combination. Suggested: to test with other MATLAB release.

Boris Reuderink - 2012-03-26 13:36:10 +0200

@Lilla, is there anythink we can still do on this bug? Otherwise I would like to close this bug, since we can't do much about it.

Lilla Magyari - 2012-03-26 14:41:30 +0200

(In reply to comment #16) hey, the problem has been solved already. It seems that a certain Linux version in combination with a certain Matlab version causing FieldTrip to crash. But this particular matlab version is relatively old (Matlab2008b), so I do not think anything should be done about it. On the other hand, it would be still necessary to facilitate the usage of newest FT/Matlab versions at the MPI in the same way as it is done in the DCCN. Lilla

Boris Reuderink - 2012-03-28 10:48:50 +0200

That is good to hear. I close this issue as RESOLVED:WONTFIX.

Boris Reuderink - 2012-06-20 15:03:20 +0200

Changed my resolved bugs to closed.