Strengthen need to have capitalised ID space prefixes
Closed this issue ยท 8 comments
We just admitted an ontology #2461 which has a lower case letter in the ID space. Since we do have a few mixed case ID prefixes this is not a disaster, but moving forward we should make sure that all new idspaces are simply capitalised, to make it easier for tools to process them. I think we should:
- Strengthen https://obofoundry.org/principles/fp-003-uris.html#recommendations-and-requirements which says "should generally be capitalized" to "all newly requested prefixes MUST be capitalised.
- Add a note to this to the Ontology reviewer docs.
Personally, I prefer that we require upper case moving forward. I creates less hassle (IMHO).
In favour of simplification wherever we can, and caps IDs would be a small gesture in this way in uniformity.
Human readability was also discussed on the call.
One note is that on the OBO Foundry main page and on the OBO Dashboard, all OBO prefixes are shown in all lowercase. It would seem to argue against the readability position for mixed case, that no one has complained about uniformity in case on OBO websites. Of course, the issue is about IDs made using the prefix, so the website consideration is not direclty relevant, but it does indicate both precedent for uniform case and a lack of reported problems with it.
My preference is also all upper or all lower. I can't remember where but I have had to write around this before, which is not a problem, just slightly annoying :)
Very important note towards @hoganwr comment above: there are two distinct concepts in OBO:
- Preferred prefixes for CURIEs.
http://purl.obolibrary.org/obo/NCBITaxon_123
is the IRI,NCBITaxon:123
is the CURIE, andNCBITaxon
is the prefix. - Ontology ids. These are always lowercase. An
ontology id
in this case is a string that allows us to look-up an ontology on the obofoundry website, for example, or the "ontology id part" of the purl string, like:ncbitaxon
inhttp://purl.obolibrary.org/obo/ncbitaxon/subsets/taxonslim.owl
orhttps://obofoundry.org/ontology/ncbitaxon.html
.
This situation is of course very confusing. Truth be told, the ideal, cleanest solution would be to make absolutely everything lowercase, but that ship has sailed (by sailed I mean "the effort of changing that infrastructure is in no relation to the benefit on the other end). If we start mandating lowercase prefixes from new ontologies, our tools like ODK need to be updated, a completely new thing is being introduced that was not there before (i.e. http://purl.obolibrary.org/obo/ncbitaxon_123
- we did have upper and mixed case, but never lower case in PURLs).
I will include this option in the vote, but I am not volunteering to enacting that change (everything lowercase) if it then gets the most votes!
Vote
๐: I think we should mandate capitalized prefixes for new ontologies. This is in line with some of the current assumptions of standard tools (ODK etc), it makes it easier to "guess" the correct capitalisation (less "hassle"), and creates "some level of uniformity", at least for future ontologies.
๐: I think we should not mandate capitalized prefixes and allows mixed case prefixes like NCBITaxon. This is in line with historical decisions (there are a number of mixed case prefixes in OBO already) and this makes prefixes more readable for humans (example "_Af_rican _P_opulation _O_ntology" --> AfPO).
๐ : I think we should, since we are creating a new rule, go all in, despite the large effort involved and the lack of a precedence, and mandate an all lowercase solution for future submissions. This will unify the concepts of "ontology ID" and "CURIE prefix".
The vote will be closed on 31th of May, 2024.