First cut reflection of work in progress for the upcoming VIVO 1.8 release

Projected Release Timing

One outcome of the meeting of VIVO sponsors at the March, 2014 DuraSpace Summit even in Washington was a request to move to a regular release schedule where timing would be predictable and new features included only if ready.  Any feature not on target to be finished by the code freeze data would have to be deferred until the next release.

A schedule of two releases per year targets a May code freeze for a June release (VIVO 1.7), and a November code freeze for December release.

The VIVO 1.8 release is being delayed to allow completion of a task that has been deemed essential for moving the VIVO project from incubator to full project status – replacement of the Pellet reasoner (VIVO-778).

Release Mechanics

The VIVO project uses the DuraSpace JIRA issue tracking system to manage issues and time tracking leading up to any given release.  Issues are categorized as bug fixes, tasks, code tasks, or improvements, grouped into components such as ontology or testing or technical debt, and assigned a status based on criticality to the release.

View unresolved and recently-modified issues for the v1.8 release. Note that issues are typically written as specifically as possible to provide clear requirements to developers and allow for realistic time tracking; when it makes sense, more detailed tasks are grouped as sub-tasks of a more general summary task.

When all issues marked with a status of blocker have been resolved, a code freeze is declared and a release candidate is prepared to allow Selenium (acceptance) testing to begin. Most releases include require several iterations of release candidates and testing to address problems detected in testing.

Selected JIRA issues:

Notes as of 12/16/14

• recently made improvements to startup time after the first-time startup (saving approximately 20 seconds)
• modularity improvements
  • search indexing (VIVO-869 being the parent issue)
  • reasoner (VIVO-778)
  • triple store (RDFService) (VIVO-225)
• VIVO-909 and VIVO-907 – if change one of the ontology files, the entire reasoning has to be re-done – the TBox reasoning that takes 20 seconds triggers a full ABox re-inferencing
  • pointed out by Eliza in testing with Virtuoso
  • a maxCardinality property was being specified as a non-negative integer but was stored as an integer, so on startup the TBox reasoner interpreted an ontology change
    • saw the same thing in Virtuoso and TDB, but doesn't show the problem with SDB
      • got around it with a kluge to the Virtuoso driver – would like to distribute with that patch if can't get around it
    • worth a couple more hours to pursue – may be an issue in the general RDFService code, that is not used for SDB
  • steps to address this may also shave more time off the startup time
• We will also need to distribute a patch for a SPARQL query syntax that Virtuoso doesn't accept

VIVO application code

For the complete list of issues under consideration for VIVO v1.8, view the full, version-specific report in Jira.

Other general goals

• Modularization
  • Major pieces of the application are selected at startup, based on a configuration file.
• Improvements in display
• Model handling
  • Technical debt: all access of triples goes through two parallel sets of channels (content and configuration)

Ontology

Documentation

• Wiki restructuring
• Customization guide

Testing