Use Case: reference objects from a Crate

[CaroleGoble]

As a data manager, I want to reference datasets and other objects by their identifiers so that I do not have to have them as encapsulated files when they are large or stored elsewhere or not mine to bundle.

The whole point of ROs was that they are collections of references - to use identifiers. To make a FAIR RO-Crate (or even a Workflow RO-Crate) means you have to fly in the face of the spec which expects to be file based and encapsulated. So the RO-Crate is just a file packaging system and not a RO system at all until it permits rather than discourages references to identifers (first rule of FAIR. reference identifiers). See Mark Wilkinson's RO-Crates for example. And the Workflow Hub examples for that matter.

[markwilkinson]

Agreed. Crates need to be identifier-centric, not file-system centric. Many datasets cannot be moved (for privacy reasons, for license reasons, for size reasons). Having a filesystem-centric standard is just... Counterproductive...

I don't want my root ID to be './'!! I want it to be the identifier of my Metadata! (As is required by FAIR!!!)

[jmfernandez]

Totally agree with you two!!!

[markwilkinson]

(and by the way... The use of PATH-oriented @id's in the Crate JSON-LD is apparently valid for JSON, but is totally rejected by the JSON-LD -> [insert other RDF format here] transformation libraries... So it's really not very useful outside of the Crates world...)

[ptsefton]

What happens in JSON-LD to RDF that's a problem?

…

On Wed, 8 Apr 2020 at 06:51, Mark Wilkinson ***@***.***> wrote: (and by the way... The use of PATH-oriented @id <https://github.com/id>'s in the Crate JSON-LD is apparently valid for JSON, but is totally rejected by the JSON-LD -> [insert other RDF format here] transformation libraries... So it's really not very useful outside of the Crates world...) — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#72 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAFYTWFN75TDUR6LK2TGH4LRLOG35ANCNFSM4MDFADEA> .

-- Peter Sefton +61410326955 [email protected] http://ptsefton.com Gmail, Twitter & Skype name: ptsefton

[markwilkinson]

"./" and other PATH-style references are not valid identifiers in other serializations of RDF, apparently (or at least, the libraries reject it, even if it is in-spec)

[stain]

I've addressed this issue in #74, and how the RDF parsing issues, caused mainly by @base: null in #73