Ontologies CURIE or URI?

From @sveinugu:

For fields with ontology contents, you specify that the format should be CURIE (e.g. “OBI:0000626”), requiring the use of a CURIE lookup service, say “identifiers.org” or N2T.net. However, as good as all ontologies make use of persistent, globally unique IRIs as term IDs (to be compatible with linked data), I think the data type for ontology terms should instead be an IRI (e.g. http://purl.obolibrary.org/obo/OBI_0000626)

For other fields, such as “sample_id” and “study_id”, it makes more sense to use CURIE ids, however these are only specified to be strings. Since the gold standard for identifiers is persistent IRIs, one should still probably support this, however, in most bases a CURIE is more available, with a simple string as a fallback for identifiers that are not automatically resolvable.

No files, tests, or entry points are named. Start by reviewing the ontology and identifier field definitions in the project specification; done means the project has resolved whether these fields should use IRIs, CURIEs, or strings and recorded the decision.

Written by the indexing model from the issue text.