updating to latest repronim locations/images

[satra]

@mih - it would be great if you could check this before i hit merge.

this is setup to:

1. use the repronim images (which should also be in the VM)
2. download the extra files from this repo.
3. add a configurable option for working dir.
4. set the call format that should work both on my cluster and in the VM.

i can't actually test on the VM since i can't download it with my current connection and the remote execution is not working well my x2go client.

[satra]

@mih - so i run the above script and it finishes and then i try to repeat the last step and i run into the impossible state error:

$ bash fsl_glm_w_amazing_datalad.sh
... # runs to completion
$ cd demo/glm_analysis/
$ datalad run singularity exec -B /tmp --bind /nobackup/scratch/Fri/satra/training/test-exercises/demo/glm_analysis/. -H /nobackup/scratch/Fri/satra/training/test-exercises/demo/glm_analysis/. .datalad/environments/fsl/image feat 'sub-02/1stlvl_design.fsf'
run(impossible): /nobackup/scratch/Fri/satra/training/test-exercises/demo/glm_analysis (dataset) [unsaved modifications present, cannot detect changes by command]

[mih]

If you ran that script git status should reveal the exported bash script as an untracked file. Just git clean -f and it should run.

I will later try if datalad rerun has the desired effect too.

[mih]

I can confirm that a datalad rerun --since HEAD~3 works as advertised.

Maybe we should set

set fmri(overwrite_yn) 1

to prevent FSL from adding a '+' directory on rerun. That way people could see straight with Git that nothing changes.

[satra]

that might be good for this exercise - we can also use that to show how a mistake in an onset file can be tracked.

[mih]

I tried, but FSL insists on renaming the folder :(

[mih]

We could revert the analysis output commit and rerun...