search

ropensci / emld

:package: JSON-LD representation of EML
https://docs.ropensci.org/emld
Other
13 stars 6 forks source link

additional validation errors #48

Closed mbjones closed 4 years ago

mbjones commented 4 years ago

I found additional validation errors in addition to #46.

1) The packageId is not used as an identifier for checking uniqueness with other id values and for resolving references. 2) Errors are thrown for annotation elements in additionalMetadata because their parent doesn't contain an id. This should be ok, because in additionalMetadata the annotation subject is determined via the describes fields. 3) The annotation/@references attribute values are not checked during validation to be sure they resolve to an id -- only the //references elements are checked.

I've begun implementing fixes for these in my fork here: https://github.com/mbjones/emld/tree/46-validation-errors

mbjones commented 4 years ago

Fixes checked in PR #49 .

cboettig commented 4 years ago

Closed by #49

Please let me know if now would be a good time to prepare a CRAN release for these changes? Or shall we hold tight and see if there's any remaining validation issues we can squash first?

@mbjones would you be willing to send another PR with updates to NEWS.md describing the validator changes? Also, the DESCRIPTION author list is now out of date -- may I add you and @amoeba as co-authors there?

mbjones commented 4 years ago

I think another release now would be good -- those were serious errors and we should get them out into the community so that there's less confusion over correct validation.

I'll send another PR and update both files, and I'll bump the version to 0.4.0 in DESCRIPTION. Do you want to run codemetar again or just update the codemeta.json file manually with the additional info?

cboettig commented 4 years ago

Thanks!

Feel free either way on the codemeta, I run the codemetar script at submission anyway.