Thanks Neil,
There is a user preference option, introduced so that users might define their own atlases. I doubt it gets used much. It is editable in the preferences dialog… that FSLview doesn't detect that there are mistakes in that preference is clearly sub-optimal. I will add this to the bug list.
The full path "fsldir" was actually intended again to solve an infrequent issue where the user's environment isn't FSL friendly.
Infrequently used bits of code lead to bugs and disaster… I should know better.
Sorry,
Dave
On 20 May 2013, at 07:36, Neil Killeen <[log in to unmask]> wrote:
> Hi Mark
>
> so I think we have solved it. It was multi-factorial. A combination I believe of permission corruptions
> but also an odd file sitting in the user's .qt directory
>
> We found .qt/fslrc
>
> with these contents
>
> atlaspath=/usr/local/fsl/5.0.1/data/atlases^e
> fsldir=/usr/local/fsl/5.0
> mni=/usr/local/fsl/5.0/data/standard/MNI152_T1_2mm_brain.nii.gz
>
>
>
> 1. What writes this ? Not all users I looked at had the fslrc file. Many had other harmless files.
> 2. It's bad that the full atlas path has been translated (as when installations change the root e.g. /usr/local/fsl/5.0
> does not change as it's symlinked to the current actual install e.g. /usr/local/fsl/5.0.1
>
> That ^e on the end of the first line looks interesting...
>
>
> regards
> Neil
>
--
Dave Flitney
Oxford Centre for Functional MRI of the Brain
E:[log in to unmask] W:+44-1865-222713 F:+44-1865-222717
URL: http://www.fmrib.ox.ac.uk/~flitney
|