PCDM 2.0

[escowles]

For tracking proposed changes to PCDM in the wiki page:

https://github.com/duraspace/pcdm/wiki/PCDM-2.0

[azaroth42]

:+1: otherwise FileSets can have members and files and relatedObjects, as a subclass of Object. None of which we want.

[tpendragon]

So I'm working on BookConcerns right now, and am starting to think about moving over Plum's structure editor, but to do that I'd like some level of agreement on what Logical Orders look like. I'm happy to be overridden here, but my complaints are described above: The implementation of "a logical order is a related object that has all the same members, but also sub-resources for chapters, done with proxies" is expensive and hard to maintain. Things like deleting a file set quickly becomes nasty to sync up.

Can we do better, or is matching the ordering and structure for logical structure important enough to pay the price?

Edit: Another thing - can we somehow build in the fact that items shouldn't be re-orderable by the logical structure into the data model? Or maybe it should be re-orderable, but that feels odd.

[escowles]

@tpendragon: is this a modeling concern, or an LDP implementation concern? And does having (Top)Range be a subclass of pcdm:Object imply that we'd implement it the same way in LDP?

IMO, it would be fine to implement (Top)Range as a single LDP resource with hash URI children, instead of IndirectContainers with Proxies, etc. And I'd be happy to reconsider whether these should be subclasses of ore:Aggregation instead of pcdm:Object if that is a barrier to this implementation strategy.

[escowles]

Based on mailing list discussion, I'm closing this omnibus issue and creating separate issues for the main topics of discussion:

• 59 Add a FileSet class

• 60 Add AlternateOrder, TopRange and Range classes

• 61 Stick with RDFS or switch to OWL

• 62 Remove the Work class from the Works extension