Generic IDs in HQDM Core Types

[twalmsley]

At present, this HQDM implementation uses String as the ID type which means that the RDF code has to convert Strings to and from IRI objects, which not only looks messy but means a lot of new objects get created when they might not be needed.

It is possible to parameterise the core types along the lines of Thing<I> where I is the type to use for Thing IDs, but:

• This will impact all classes and interfaces in the HQDM model and impl packages.
• It will mean all client code has to provide the type parameter when referring to these core types.
• All of the object creation services and RDF object creation services will be affected.
• There might be unforeseen problems for clients of MagmaCore, and it might be a breaking change for some clients.

This would probably take 2 to 3 days and would be a very large pull request for someone to review, then all clients would need to change their code because they currently expect String IDs and would now have IRI IDs instead.

I might be able to get this done as part of one of my projects, so I'll make a start and see what the impact is. Feel free to add comments, especially if you see any problems or a better way to do this.

[twalmsley]

The impact of this change is huge, it affects:

• Every interface in the model package.
• Every class in the model.impl package.
• All of the builder classes.
• All of the core module classes.
• All of the examples.
• All client code will be broken.

Is it still worth it? See branch GH129-generic-thing-ids for an attempt at the changes - I didn't get round to fixing the builder classes once I saw the full scope of the changes needed.

[twalmsley]

My conclusion is to use IRI objects for entity IDs and to merge the hqdm module with hqdm-rdf. This has less impact than using generic IDs but is still a breaking change.