Oscar Esteban

> Which seems more worthwhile, updating the minimum nibabel I think to update nibabel's minimum version is the right call here.

Okay, I'll edit the title -- I think I should have a second look after the long break I've taken with the implementation.

The overall idea seems very reasonable. At a first sight, I'm inclined towards a language reusing `--ignore`, e.g., `--ignore confounds-dvars confounds-fd confounds-{a,f,}compcor` Alternatively, we could be more lenient handling errors...

Hi Craig, that would be the other facet of the discussion in poldracklab/fmriprep#747. As Chis mentioned, we will soon rethink the treatment of fieldmaps in FMRIPREP.

This could be done within this function: https://github.com/poldracklab/sdcflows/blob/fcc9ee0d242aaf52e460afa80b5c7c1bfa1a8c63/sdcflows/workflows/base.py#L260-L292

That's a good idea, but how would you check that from the BIDS input?

Yep, the problem is that the heuristics to assign fieldmaps and target data without the `IntendedFor` are quite complex and subject to opinion (e.g. the comments about shimming above). This...

After thinking this through, and having bids-standard/bids-specification#622 in the works, I think this issue can be moved back to fMRIPrep.

I think your target here would be writing an equivalent to `init_single_subject_wf`: https://github.com/poldracklab/fmriprep/blob/a6fe5500ad2e49fb57dd6a4c98155be4a41570bd/fmriprep/workflows/base.py#L210-L462 where no anatomical workflows are instantiated and run. Without meaning to discourage you, I must note that...

They are set in by connecting this node: https://github.com/poldracklab/fmriprep/blob/a6fe5500ad2e49fb57dd6a4c98155be4a41570bd/fmriprep/workflows/base.py#L363-L364 to those fields in the top-level workflow, eg.: https://github.com/poldracklab/fmriprep/blob/a6fe5500ad2e49fb57dd6a4c98155be4a41570bd/fmriprep/workflows/base.py#L407-L408 sets the anatomical inputs.