...
- _ Ryan, Excellent Point, what we are seeing today in both LoC and DCMI communities is further adoption of RDF to provide representations of complex structured metadata. So for instance, your FoaF "value" would actually be modeled as a set of explicit RDF Statements. This is where applying the concepts behind RDF Models/Graphs allows us to make such attachments. See also: DSpace+2.0-Modelling_DSpace_Objects. Taking this a example a step further to express how this would be supported, we would have a strong example of modeling and embedding Structured metadata in RDF... --Mark Diggory 14:40, 23 September 2008 (EDT)_
...
- _ I think this should be taken a step further in that its the responsibility of the underlying implementation of a Content Management Service to address the canonical identifier it uses in the generation of its Content objects. This means that the underlying CM Service implementation is also an ObjectIdentifer service capable of minting new identifiers when required when generating objects in the repository. To give key examples (a JCR-170 store may randomly generate unique ids used to identify its contents, a Fedora store may generate info:fedora/..., urn:uuid:..., or other identifiers that represent its contents, and our default implementation will definitely be generating urn:uuid:... identifiers. I currently see in the trunk that DSpaceObject has a "setIdentifier" method exposed on it, I think this should never be the case and that DSO retrieved from the Content Repository have this set and fixed in their construction as part of the Objects creation before being delivered tot he user. Identifier objects should be immutable._ --Mark Diggory 16:02, 24 September 2008 (EDT)
- Also consider how to preserve the persistent identifiers that are already associated with a content object at the time it is ingested – for example, Items transferred from or replicated for another repository. It's fine to assign new identifiers too, especially if they are only for internal (to the DSpace site) use, but the must also be findable through its old persistent identifiers. LarryStone 02:38, 8 October 2008 (EDT)
...