Sprint Team

General

Meetings

4/21

4/22

  • Kevin
  • Eric
    • Working on Event audit design
      • https://www.pivotaltracker.com/story/show/69811150
      • OWL ontology covers 80% of what we need to describe events
      • Events that Fedora/JCR generates don't match 1:1 to PREMIS. Might need something other than event machinery. 
        • Add new events? Attach PREMIS generation to methods in the code? 
        • Will post a proposal to the wiki
    • Will help Kevin with Tomcat Roles PrincipalProvider ticket

4/23

4/24

4/25

4/28

  • Kevin
  • Eric
    • Working on Javadoc build warnings
    • Also working on: Test implementation for allowing referencable auth nodes
      • https://www.pivotaltracker.com/story/show/70111766
      • Question from Andrew: Does this approach intersect with the UNC XACML work?
        • It's an alternate to the XACML implementation, when fine-grained XACML isn't needed
      • Question from Andrew: You enable writing authorization rules in system space. In the existing pattern, authorization rules are determined in hierarchy of content nodes. Is there a use case for the new approach? Can we support both models?
        • Possible, but why would you want to keep them embedded in the node? The purpose of the new model is reusability.
        • We should think about whether or not there is a strong use case here.
    • Event audit design
      • https://www.pivotaltracker.com/story/show/69811150
      • Question from Andrew: How will we actually store the events on an ongoing basis? How will we retrieve event information from the repo? Use fcrepo transform?
        • PREMIS kept in container (child) node and use the OWL ontology as RDF properties on the PREMIS container node.

4/29

  • Kevin
    • Working on ExecutionError while trying to ingest 500k records

      • https://www.pivotaltracker.com/story/show/69454152

      • Definitely an uptick in ingest time as more objects are added

      • More data required for an accurate assessment

      • Need to set a large target (1 million, 10 million, etc.) to test performance with very large data sets
  • Eric
    • Working on Test implementation for allowing referencable auth nodes

4/30

 

5/01

  • No labels