allowing an optional LICENSE file, next to README and CHANGES

[sappelhoff]

In BIDS we do not make any strong requirements or recommendations about licenses to use for the data. We have a very short appendix on licenses and we ask users to specify the license in a string in the dataset_description.json file.

I think it'd be reasonable to allow an optional LICENSE file in the same vein as the README and CHANGES.

Reasons:

In the review to get the EEG-BIDS paper published with the BIDS example datasets, a reviewer required a LICENSE file ... i.e., the string in the dataset_description.json file was not enough. So I added a LICENSE to the example dataset and then had to add also a .bidsignore, because we don't allow a LICENSE file.

Thus my reasons for requesting a LICENSE file are ...

• it's a low cost feature
• it's probably required relatively often and is widely used
• people do it anyways but currently have to put a .bidsignore into the dataset additionally

Possible issues I can imagine:

• LICENSE file and dataset_description.json field could be in conflict (due to user error)

[satra]

a reviewer required a LICENSE file

did the reviewer indicate a specific reason for this? Some licenses do require a specific statement to be added, but not necessarily a separate file.

[sappelhoff]

I just looked into the old emails again the exact reasoning was a bit different:

• I had an ODBL license (https://opendatacommons.org/licenses/odbl/summary/)
• On the Open Science Framework, I can't select this license from the dropdown menu, so I picked "other"
• the reviewer just saw the "other" and didn't know that there is a "license" field in dataset_description.json

Then I got this reply:

ideally this [the license] would be stated on the landing page of the data record, it currently lists ‘Other’. If this is not possible, we would encourage you to upload a separate Licence file next to the Read Me file at OSF to highlight this.

[effigies]

This makes sense to me. Validation for consistency can probably check that the license name appears somewhere in the license file, and assume the curator isn't intentionally deceptive.

[sappelhoff]

during the maintainers meeting another reason to allow LICENSE files came up:

--> some licenses require that the full license text is shipped with whatever is being licensed that way.