Attendees

• Greg Jansen
• Unknown User (escowles@ucsd.edu) 
• Michael Durbin
• Eric James
• frank asseg
• Ye Cao
• Matthias Walter
• Markus Haarländer

General

• Indicates who took minutes - 
• Call-in: Google-hangout at: 
  
  • https://plus.google.com/hangouts/_/event/ccnv0ba94h3fb2je3gscunqc49g

Agenda

1. Sprint b3 planning
   
   1. AuthN/Z
   2. Triplestore
   3. 3-4 Upgrade
   4. Large files
   5. Clustering
   6. Search and/or Sequencers

Minutes

1. Project and process introduction
   
   1. Pivotal tracker
      
      1. Tickets, points, phases, etc
   2. Git flow
      
      1. Working branches, submit pull requests
   3. Unit and integration testing
      
      1. Ensure minimum of 75% unit test coverage
      2. Sonar
2. Ticket walk-through and assignments
   1. Esme
      
      1. Working on triplestore tickets, starting with use cases and sample data/queries.
      2. Documenting triplestore deployment, then how a user can create objects and inspect results in a triplestore.
   2. Michael
      
      1. Scalability issue with nodes that have many children (e.g., fcrepo3 repository with a million objects).  All children are loaded and included in the REST API response.  Need broader discussion of how to address this (using paging, simulated hierarchy, etc.).
   3. Eric
      
      1. Working on large files, esp. whether performance of SHA1 checksumming can be improved by using native OpenSSL call instead of Java.
      2. Wiki documentation of large file ingestion.
      3. Testing whether content added to a projection using the REST API is stored in the projection, or in the repository.
   4. Matthias/Ye/Markus
      
      1. Need to review existing indexing work and develop design document for searching/indexing.
         
         1. Design - Customizable Search Index
      2. Goal is customizable, configuration process, need to identify what can be configured.
      3. MODS sequencer: should be a more generic XML sequencer, wait for indexing design document.
   5. Greg
      
      1. Access control models to attach roles to fedora nodes.  Goal is to add simple policy enforcement in Java (not delegating to external system).
      2. After that: Adding XACML to a separate workspace and link them to parts of the repository to define policy.

Actions

1. All to flush out details of respective tickets
2. Setup meeting to discuss scalability of nodes with many children (Greg Jansen)