Facilitator: Justin Coyne, DCE
Note taker: Andrew Myers, WGBH
- What needs to happen to AF to work with F4?
- Rights metadata
- How can Samvera better leverage F4 features?
- What are we doing to move toward really using F4 and getting away from F3?
- Figuring out direction of community re: embracing LDP (didn't really get to this, ran out of time)
What needs to happen to AF to work with F4?
For an MVP (minimum viable product):
Main effort is coming from Sufia folks, so the bullet list is currently "whatever is needed for Sufia to work."
Other things to address. Maybe put these on roadmap for point releases...
- Do we want to implement weak references?
- Do we want to make datastreams go away?
- Ability to set arbitrary properties on datastreams.
Current trajectory looks like Hydra 7 will be for Fedora 3 only. Hydra 8 will be for Fedora 4 only.
Is this a good promise to make? Still discussing that, but we should be clear with whatever we decide.
Absent resources, it's going to be tough to develop code to make AF8 work with F3. And if you do have those resources, should probably consider putting them toward upgrading to F4.
It might be a good time in Hydra 8 point releases to re-think the model layer, and explore other patterns, e.g. data mapper instead of active record.
Strategies Moving from Fedora 3 to Fedora 4
There was once a "connector API" but that doesn't work anymore.
UVA says they are committed to helping with migration strategies.
Projection strategy: Use projection of Fedora 4 to project over your old Fedora 3 repo.
Hydra 8 alpha continues to use the same
Should we convert Hydra 7 rights to RDF?
We would need predicates.
Right now there is an issue with having rights metadata on the Hydra level, as opposed to Fedora level, in that we don't have any access metadata when look directly at Fedora objects (i.e. from other systems).
Do we want to make it capable to put rights metadata into Fedora, so you get your access control info from Fedora?
How can Hydra better leverage F4 features?
There is currently an issue for supporting this.
Should probably hold off on putting it on the roadmap for Hydra 8, to prevent blocking that release.
Fedora has its own Solr indexing workflow. Is there interest in having Hydra take better advantage of that?
Yes, we think so. It could allow AF to shed some indexing functionality. Needs more exploration.
We should be conscious to use LDP language and terms when called for in ActiveFedora (as opposed to ModeShape language).