Currently, many things are tied to the working directory and the output directory during execution. Particularly when deploying this on CBRAIN, which has a particular way of working with file structures, I've had to kludge things.
After a discussion with Munira, it makes more sense to have an intermediates directory as an argument, allowing re-use of calculated values (which saves TONS of time) and a separate output file specification.
I think this will make usage clearer in general and will be wonderful for the boutiques descriptor and making it work nicer under our version of CBRAIN.
I will look at some approaches from other tools/descriptors to make something I think will be easiest.
Currently, many things are tied to the working directory and the output directory during execution. Particularly when deploying this on CBRAIN, which has a particular way of working with file structures, I've had to kludge things.
After a discussion with Munira, it makes more sense to have an intermediates directory as an argument, allowing re-use of calculated values (which saves TONS of time) and a separate output file specification.
I think this will make usage clearer in general and will be wonderful for the boutiques descriptor and making it work nicer under our version of CBRAIN.
I will look at some approaches from other tools/descriptors to make something I think will be easiest.