[FIX] Add coordsystem-specific definition of DigitizedHeadPoints

[tsalo]

Closes #1022.

Changes proposed:

• Add definition for DigitizedHeadPoints for coordsystem files.

[tsalo]

@Moo-Marc would you mind taking a look at the proposed changes?

[Moo-Marc]

Thanks,
The contents looks good, though I'm not familiar with the syntax.

[tsalo]

@Moo-Marc thanks for taking a look. Here is the updated definition, for posterity:

@sappelhoff do you know what kind of relative path DigitizedHeadPoints should be? I put participant_relative, like IntendedFor, but I don't know if that's correct. The definition is a little vague.

[sappelhoff]

good question, I don't know :-(

in this example, it's apparently relative to the coordsystem.json file itself (where the DigitizedHeadPoints metadata is specified): https://github.com/bids-standard/bids-examples/blob/df982afc719751d4b7f38d426aa38fc6832e06fa/ds000117/sub-01/ses-meg/meg/sub-01_ses-meg_coordsystem.json#L8

that looks super odd and is probably a bug. Another reason why we need #918

Maybe @Moo-Marc knows? Or @robertoostenveld?

[Moo-Marc]

I was a bit confused about paths as well and following the examples. So for this I used the file name only (it's in the same folder). I like the idea of #918.

[robertoostenveld]

I suspect that back in the days of writing the MEG specification the idea was to have this similar as IntendedFor with a past specification that is relative to the JSON file that specifies it. So an example without explicit path like "sub-01_headshape.pos" would be in the same directory as the coordinatesystem.json

[sappelhoff]

I suspect that back in the days of writing the MEG specification the idea was to have this similar as IntendedFor with a past specification that is relative to the JSON file that specifies it.

Thanks Robert, I think you are right.

@tsalo would that mean that we need a new "format" for paths? Currently we just have dataset_relative and participant_relative:

bids-specification/src/schema/README.md

Lines 219 to 221 in 67c3d90

	- `dataset_relative` (relative paths from dataset root),
	- `participant_relative` (relative paths from participant folder).

of course, all of these will be DEPRECATED anyhow once we merge #918 ... but we still need to solve this issue for backward compatibility

[tsalo]

@sappelhoff I think you're right. Since it should be relative to the file in which the field is present, does file_relative sound good?

[sappelhoff]

does file_relative sound good?

👍 sounds good to me!

[tsalo]

It's now set to file_relative.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 71.50%. Comparing base (ee9e8b5) to head (97c75b8).
Report is 1426 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1023      +/-   ##
==========================================
+ Coverage   70.53%   71.50%   +0.96%     
==========================================
  Files           9        9              
  Lines         930      930              
==========================================
+ Hits          656      665       +9     
+ Misses        274      265       -9     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[sappelhoff]

I think this is good to go then. Thanks a bunch @tsalo (and @Moo-Marc for noticing)