[Nipy-devel] [nipype] Bet output '_bet' -> '_brain'

Dav Clark davclark@berkeley....
Wed Jan 6 20:15:46 CST 2010


On Jan 6, 2010, at 5:48 PM, Satrajit Ghosh wrote:

> hi dav,
>
> there is one option currently that could help in this situation.
>
> pipeline
> .run_in_series(updatehash=True,force_execute=['nameofbetnode'])
>
> this will rerun bet only, but presumably since it will generate the  
> exact same file, no other dependent nodes will be run.

I don't want to run bet at all! It already ran, I just want to rename  
the output files (perhaps even manually).

> however a better option might be to add something like a noderepair  
> option to run_in_series.
>
> noderepair : a dict containing repair function handles
>     e.g noderepair['nameofbetnode'] = repairfunc
>
> this function can then be executed in the proper directory right  
> before run() or aggregate_outputs() is called.
>
> what do you think? this is very easy for me to add.

I don't want the facility to run a function like this... it seems a  
bit counter to the spirit of the project. I.e. - it could leave the  
output in a state that doesn't truly reflect the specification in the  
graph.

My issue is that I already ran everything through data-processing- 
wise, but the rename of a file there will trigger a re-run (which is  
annoying). I just want to rename the file in the pipeline spec graph,  
and rename the files on the filesystem, and then voila, everything is  
OK.

Certainly, I could do this "by hand" in a fairly cumbersome way.

Thanks!
Dav



More information about the Nipy-devel mailing list