Change arg order to make `kind` a required field about mne-bids HOT 6 CLOSED

As for differences between EEG vs MEG, won't they be ironed out before the extensions are merged into the main spec?

However, there will probably always be slight differences, such as different coordinates.json files between MEG, EEG, and iEEG. --> a "kind" argument would allow us to handle these very easily and explicitly.

from mne-bids.

Could we not infer it directly from the info object?

from mne-bids.

we use it internally to make sure we are pulling modality-specific channels and info. I think inferring would be problematic with files with combined eegand meg. also, there are some differences in the json writing for meg versus eeg so I think it would be better to remain explicit

from mne-bids.

If I remember correctly, the kind argument was introduced by @choldgraf to deal with the fact that the iEEG validator was not yet merged into the main BIDS validator. It allows us to keep the check against the MEG validator in the tests while giving users the option to try the experimental iEEG conversion of MNE-BIDS. Once the iEEG validator is merged, we should remove the kind argument so that both the MEG as well as the iEEG conversion is tested.

As for differences between EEG vs MEG, won't they be ironed out before the extensions are merged into the main spec? Basically I feel the only spec we should respect is the main spec. The extensions are just temporary ...

from mne-bids.

as well as we are able to swap among the metadata files appropriately and it works seamless, i would be OK with the kind argument going away. we would need to decide before release otherwise we will be breaking api 🙃

from mne-bids.

@teonbrooks let's close this issue as it's a duplicate of #73 and #105 ?

from mne-bids.