Attendees
- Andrew Woods
- Kevin S. Clarke
- Scott Prater
- Greg Jansen
- Unknown User (escowles@ucsd.edu)
- Longshou Situ
- Michael Durbin
- David Wilcox
- Unknown User (bbpennel)
General
- Indicates who took minutes -
- Call-in: Google-hangout at:
Goals
Put touches on for Beta release
- Refactor AuthZ and implement for XACML
- Enable transparent JCR hierarchy
- Demonstrate pluggable PID generator
- Demonstrate sequencer
- Determine performance characteristics
- Remove JCR leakage in REST-API
- Clean-up logging
- Clean-up code styling
- Bring unit and integration metrics above 80%
- Explore options for "transparent" storage
- Explore options for clustering for scalability
Next Sprint
- Clean-up UI
- Clean-up documentation
Minutes
- Goals
- Emse: Clean up logging output so we don’t scare new users
- Don’t output stack trace when we have code to address the issue (e.g. 404)
- Greg: What are we doing to refactor authorization?
- Existing authorization framework was refactored last sprint so content nodes reference ACLs stored in system space (rather than hanging off the nodes themselves)
- Mike: Clean up JCR leaks
- General vote to de-prioritize UI issues until next sprint
- Emse: Clean up logging output so we don’t scare new users
- Tickets
- Ensure a clean shutdown of the repository on container shutdown.
- Write an integration test to ensure that transactions cannot be hijacked.
- https://www.pivotaltracker.com/story/show/66014686
- Will be done soon.
- Determine potential of horizontal scaling
- https://www.pivotaltracker.com/story/show/69178852
- Ongoing conversation with ModeShape community
- ExecutionError while trying to ingest 500k records
- https://www.pivotaltracker.com/story/show/69454152
- Will close this ticket after adding notes to a wiki page
- Need new ticket to specifically address loading the repository with a million objects
- Getting lock metadata multiple times in a session throws exception.
- https://www.pivotaltracker.com/story/show/70597666
- Make sure all RESTful API calls log out
- Update to Modeshape 4 / ISPN 6
- https://www.pivotaltracker.com/story/show/70387904
- Waiting on new ModeShape release
- Update upstream dependencies
- Update checkstyles for fcrepo-jms-indexer-pluggable
- Turn on checkstyle checking for fcrepo4's test classes
- https://www.pivotaltracker.com/story/show/69355194
- This will flag some errors in the current codebase
- Will need to be careful when committing this to avoid merge conflicts
- Create strawman for holding ModeShape configuration
- https://www.pivotaltracker.com/story/show/65979104
- Low priority - moving to the icebox
- Add UI "Create Lock” Button
- https://www.pivotaltracker.com/story/show/70438216
- Low priority - moving to the icebox
- Only enable AuthZ UI form when applicable
- https://www.pivotaltracker.com/story/show/63178228
- Low priority - moving to the icebox
- Hide/disable interface add object/ds option if FilesystemConnector is read-only
- https://www.pivotaltracker.com/story/show/65899330
- Low priority - moving to the icebox
- Improve and simplify fcr:versions response triples.
- https://www.pivotaltracker.com/story/show/70475058
- Need to document expectations in ticket
- AuthZ master plan?
- https://www.pivotaltracker.com/story/show/70648810
- Will be closed once Greg pulls together the tickets for working on auth
- Refactor fedoraUser AuthZ expectation
- https://www.pivotaltracker.com/story/show/70306268
- Lump all AuthZ tickets together and view them as a whole
- Enable AuthZ on Filesystem Federation
- https://www.pivotaltracker.com/story/show/68016728
- To be thrown in with authorization tickets
- Add mix:versionable mixin (if missing) when user attempts to snapshot a version of a node.
- https://www.pivotaltracker.com/story/show/70289498
- Enabling versioning introduces some overhead
- This would make sure versioning is available but only enabled when needed
- Evaluate versioning and performance
- https://www.pivotaltracker.com/story/show/70288940
- Test the impact of various versioning operation
- Versioning of container nodes should be more efficient.
- Modeshape doesn't release locks held by sessions when logout() is invoked.
- https://www.pivotaltracker.com/story/show/69574878
- One line change, just need to get test to run successfully
- Address "Critical" Sonar Issues
- https://www.pivotaltracker.com/story/show/70584818
- Scott will look at the Sonar coverage reports and create tickets for trouble areas to help coordinate multiple developers working on improving test coverage.
- Clean-up Fedora wiki main page
- https://www.pivotaltracker.com/story/show/70673612
- The wiki looks bad in different browsers
- Remove slideshows and other content that breaks the page
- Sequencer: creating REFERENCEs for associated XACML policies
- https://www.pivotaltracker.com/story/show/70671762
- Need to go into the auth bucket
- Implement a XACML policy to support metadata reader use case
- https://www.pivotaltracker.com/story/show/66627712
- Role that can not see content but can see the object’s properties
- Fix fcrepo-jms-indexer-pluggable tests
- https://www.pivotaltracker.com/story/show/70673426
- Everyone needs to see if this error is popping up for them
- FedoraNodes unit tests
- https://www.pivotaltracker.com/story/show/70645824
- Increase the unit test coverage
- Migrate fcrepo-generator-dc functionality
- https://www.pivotaltracker.com/story/show/70510376
- Small Maven module that can be moved out of the Fedora codebase
- Migrate to fcrepo-transform
- Javadoc build warnings
- https://www.pivotaltracker.com/story/show/69747108
- The same package should not be spread across different modules
- Property CRUD performance testing
- Version history documentation
- https://www.pivotaltracker.com/story/show/69744104
- Put in the icebox for next sprint
- TXNs and Filesystem Federation
- https://www.pivotaltracker.com/story/show/69444336
- Test transactions features against file system federation to make sure everything works
- Update code header copywriter
- https://www.pivotaltracker.com/story/show/69666218
- Need to update date to 2014 across all files
- Document Id Translator Pattern
- Create Maven BOMs for fcrepo dependencies
- https://www.pivotaltracker.com/story/show/63571456
- Maven 3 BOMs are a new structure for expressing dependencies. Using them should replace the fcrepo-jcr project (which is just a pom.xml for dependencies, and a dummy Java file to suppress the warning if there is a package with no source code).
- Reduce build and runtime logging verbosity
- Investigate what data is available on incoming authz requests for newly created objects, i.e. in Session
- How many objects can F4 hold?
- https://www.pivotaltracker.com/story/show/70648658
- There are several testing tickets related to ingesting large numbers of objects. We can approach these in two different ways:
- Wait for the auto-generated hierarchy updates to land.
- Create top-level containers to partition the repository so this work can go forward without waiting for the auto-generated hierarchy updates.
- Replace POST auto-hierarchy
- Employ transparent auto-hierarchy for objects
- https://www.pivotaltracker.com/story/show/70648694
- This is a strategy to avoid performance problems with large numbers of children under a single node. When a child node is created, the identifier (either supplied or auto-generated) is used to generate a hierarchy, and that hierarchy is hidden from the user. This is pretty involved and will have a broad impact across the codebase, and is important for performance. This is a good area for multiple developers to collaborate. Also, the Replace POST with auto-hierarchy ticket is related and will probably be largely addressed by this.
- Test repo export at scale
- https://www.pivotaltracker.com/story/show/70648736
- Both for testing large-scale export performance, and making sure it is functional at large scale. We should create a large number of objects, export the data, import the export file into a fresh repo, and then re-export it. The export files should match. We could develop SPARQL queries or other strategies for verifying that the repository content is loaded correctly, but it's probably more efficient to manually inspect a random sample. Longshou's sample dataset may be useful for testing this.
- Tighten code styling
- https://www.pivotaltracker.com/story/show/70648784
- Tightening code styling is a larger discussion. But a good task to work on now to provide options to discuss and act on at the Asheville hackfest.
- Move git repos to new git organization
- https://www.pivotaltracker.com/story/show/70671342
- We should move the main fcrepo4 and other active repositories out of the "futures" github organization (which isn't a very good label now). They should be moved to a new "fcrepo4" organization. We can repurpose the futures organization for a "Fedora Labs" area similar to the new Hydra Labs. We may also want a separate area to separate core fcrepo4 from extended (but still "blessed"/supported modules) -- we can discuss this in the future.
- Remote pid-minter
- Unit and Integration tests to 80%
- Determine potential of horizontal scaling
- Other notes
- All developers should estimate their tickets and give some serious thought to whether their tickets are achievable this sprint, so we can shift the load -- better to this earlier and be able to plan effectively.
