Manipulate `bids.configs` settings about fitlins HOT 11 CLOSED

Well, as long as you keep the old name as an alias for a couple versions, that shouldn't be a problem.

from fitlins.

What does loop_preproc do? This is the first I've heard of bids.configs.

I'm not opposed, in principle.

from fitlins.

Its so that bids.analysis loops over preproc files rather than raw bold files when loading variables into collections. That way you don't need the raw data, only the fmriprep output

Bids config is new, but allows you to tweak such behavior that may be too niche for a functions API

from fitlins.

Might be worth renaming that option to use_preproc_images, I think the loop part is potentially confusing.

from fitlins.

Or maybe validate_preproc_images.

from fitlins.

Oh, I see. I thought that had been handled automatically, rather than via a special config option.

That's almost certainly what we want in FitLins in general, so I don't see a need for a special option. We should just do it here. If it turns out we need to index the original dataset instead, we can add an option then.

from fitlins.

Bids config is new, but allows you to tweak such behavior that may be too niche for a functions API

I'm super wary of global hidden state. It's a great way for order of module loading to affect behavior, which is a debugging nightmare.

from fitlins.

Yeah fair, although some global parameters are useful. I'm fine renaming that option as Tal suggests, and also agree it's good default behavior for FitLins. I would be fine moving it to a API parameter to load_variables

from fitlins.

@adelavega What's needed to move FitLins to using preproc? Is it just the config option, or do I need to adjust anything else?

from fitlins.

Just the config option should be enough.

See: https://github.com/neuroscout/neuroscout-cli/blob/fec99558457c9537e3674c2e251c7664edc42f28/neuroscout_cli/commands/run.py#L53

from fitlins.

(Although I should probably rename it, as Tal pointed out)

from fitlins.