ORCiD value checksums not validated

[slint]

While there is a structural/format validation for filled-in ORCiD values (i.e. that the string follows the NNNN-NNNN-NNNN-NNNN pattern), there is no checksum verification for the ORCiD value itself, which could be additional help for mistyped/pasted values.

As an example, the ORCiD 0000-0003-2680-0055 is invalid (based on its checksum) but is accepted by the form. Correct values in this case could be 0000-0003-2680-0025 or 0000-0003-2680-005X.

[abelsiqueira]

Hi @slint, thanks for the feedback. A quick question, did this happen to you? Our usual experience is people pasting the orcid, so this issue doesn't usually happen. If you use it differently we would be very interested to know more on how you use, what is your setup, etc. Our pool of "feedbackers" might be biased.

On to the actual issue, it looks straightforward, but there is a hidden problem. All current validation is done through the schema. I am not sure, but I think it is not possible to do it in the schema. In that case we would have to add it as a special case.

I will wait for your feedback so we can figure out how to proceed. Also, let me know if you have the time to contribute.

[slint]

A quick question, did this happen to you? Our usual experience is people pasting the orcid, so this issue doesn't usually happen. If you use it differently we would be very interested to know more on how you use, what is your setup, etc. Our pool of "feedbackers" might be biased.

Not directly to me, but to a Zenodo user that generated their CITATION.cff file via the tool (I am not sure what their experience was).

We discovered the issue because later on, we parse the CITATION.cff file on Zenodo as part of our integration for publishing DOIs for GitHub releases. Basically, we're trying to parse parts of the CITATION.cff metadata to use as basic information for our record (authors, title, description, etc.), and in that case the ORCiD value for the author was skipped (because it was invalid).

I also agree that ORCiDs should be either copy-pasted or e.g. autocompleted from search to avoid input mistakes, so probably this usage is not representative.

On to the actual issue, it looks straightforward, but there is a hidden problem. All current validation is done through the schema. I am not sure, but I think it is not possible to do it in the schema. In that case we would have to add it as a special case.

I understand that based on the schema itself that can prove difficult since there is additional logic involved for computing the checksum. Even on Zenodo we're not validating ORCiD checksums on our form/frontend, but it's a more thorough heavier check performed on our REST-API backend.

So I would say this feature request is not critical, but more of a nice to have in this case.

[abelsiqueira]

Thanks for the clarification. I should do a larger public feedback session to get more examples.
I will add to a list of possible issues for our future sprints.