Time/Place

This meeting is a hybrid teleconference and IRC chat. Anyone is welcome to join...here's the info:

Attendees 

Agenda

  1. Plans for WebAccessControls?
  2. Expectations of Fedora 4 configuration/configurability
    1. Content modeling/structural validation features available in 4.0.0 (see discussion in https://www.pivotaltracker.com/n/projects/684825/stories/81698538)
  3. OAIProvider information predicate:property storage on root node
  4. ...

Minutes

  • WebAccessControls
    • Interest from the Hydra and Islandora communities in adopting WebAccessControl ontology for access control metadata
    • Synergy with LDP
    • Justin Coyne has a PR to implement WebAccessControls in the Hydra Access Controls gem
    • Next step: need to review the Hydra implementation and understand how it uses the ontology and use that to inform Fedora implementation
    • How do groups, IP ranges, etc. map into WebAccessControl: upstream system provides those in headers? resolution of IP ranges/users/etc. within Fedora?
      • IP address -> group mapping can have a lot of churn, so may not be a good fit for storing in the repository
    • Who's interested in providing use cases, testing, implementation?
  • Content modelling and validation in Fedora 4
    • Recent discussions about CND content modeling functionality: what kinds of support should Fedora 4 have for using CND types to constrain repository content?
    • Main issue is that Fedora would like to avoid leaking underlying implementation details (like JCR node types) – when people depend on the implementation, it makes the project less sustainable
    • For example, recent updates to store all values as strings with RDF type information – if people we using CND constraints on datatypes, that would be broken
    • Stefano: We've been using CND constraints and not being able to use it would create a burden to replace the validation functionality, which is a large part of the work we've done
    • Adam: bigger problem is Fedora doesn't have content-modeling functionality – a system may need to do modeling or validation, but probably not in the repository itself
      • Big differences of opinion about validation – some people want strong validation, some people want a refrigerator (accepts whatever you put in and gives it back)
    • Maybe generic RDF validation approaches could work, as a sequencer or separately?
    • There is some validation at the RDF processing level: URIs have to be well-formed, etc.
    • General direction: all properties are stored as strings at the JCR level
    • CND structural constraints should still work, but datatype validation would not work any more