At present if a requested variable (e.g. specified with -v
) is not found in any input table, there is no error or warning message given. This is undesriable as UK Biobank does depricate variables and hence variables once used can 'disappear' from the source table.
I propose that at least a warning should always be added to the log when a variable is requested but not found in any input table. Additionally, an option could be added to turn the warning into an error, but there are presently too many options :) so maybe just the warning is fine.
Tom Nichols (4b85da6c) at 13 Oct 08:49
Tom Nichols (b6b59c37) at 13 Oct 08:48
Tom Nichols (b6b59c37) at 13 Oct 06:54
Remove use of xvfb and $prefix for fsleyes
At least on MacOS Ventura, fsl_gen_3D
hangs when fsleyes
is launched to do the actual rendering. With help from @paulmc, we figured out that the hanging could be avoided if the call to fsleyes
wasn't launched via bash
(see fsl_gen_3D#L89-122).
This MR impliments that change, which now works on MacOS, though someone should probably make sure it works on Linux or other platforms.
It can be tested like
fsl_gen_3D $FSLDIR/data/standard/MNI152_T1_1mm.nii.gz out
and confirming that out_{1,2}.png
are created.
CC: @falmagro
Tom Nichols (29ec7345) at 12 Oct 20:12
Remove bash prefix for fsleyes
As I understand, the recommended practice in FSL is to always reference images with out a specific nii
, nii.gz
or img
extension, allowing the code to be agnistic to the particular flavor of NIFTI currently in use.
When useing fsl_deface
however, I got an error with fsl_gen_3D
Error: Input image is not readable.
even though the image was present and readable. The problem had been that I had specified the image without an extension, and the check for readability was done directly with the supplied image. I think the solution is simple, just deleting that block, and let any readability issues get caught by the imtest
that follows (and that is compatabible with extensionless images filenames).
CC: @falmagro
Tom Nichols (e86375ab) at 12 Oct 12:14
Now should work with imagebase nameonly (no .nii.gz)
Tom Nichols (d0d6b0c1) at 20 Sep 13:30
Merge branch 'enh/replace-minmax' into 'master'
... and 87 more commits
Having used the histogram tool intenseively in the last day or so I noticed a few quirks that, while are all very low priority, I thought I'd mention them:
That's it! Just food for thought.
Super, sounds good! Once you've made a mod I'll give it a spin, and once it's pushed into the world I'll tweet about it in response to the AFNI tweets about their recent paper.
Tom Nichols (4d98a7df) at 31 May 17:32
Tom Nichols (4d98a7df) at 30 May 13:58
Now made variable names and subject Ids optional
Tom Nichols (3c846804) at 30 May 13:34
Added dimension checks