DiscoverEd/Meetings/2010/06/21

From Creative Commons
Revision as of 17:23, 21 June 2010 by Dithyramble (talk | contribs) (add notes)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

DiscoverEd Checkin 21 June 2010

Sprint Follow-up

  • Outstanding tasks
    • Asheesh writes up a team report for his team (-:
      • structured as a spec page
    • Raffi writes up a team report for the tag-adding team
  • Issues from sprint
    • NY has been refactoring the RdfStore, and was sad to see half-finished refactorings in the codebase. Going forward, we should pay more attention to these refactorings. When we add a new helper method that simply calls an existing method, maybe we can simply replace the old method. That way we could avoid leaving both the old and new versions in the class.
  • Tests
    • running from Ant: Asheesh and Raffi will confirm that the tests in the branch "next" do pass. Nathan will push an ant target he wrote once his laptop is resuscitated. AL and RKL will then use ant to run the tests.
    • source tree separation (src/tests/... instead of src/java/...)
      • This seems to be the pattern Nathan has observed in large Java projects. It also allows you to easily create a "run-time" that excludes your testing code.

What we were working on before the sprint:

Excluding a curator from a search

  • We had written a large test, and were in the process of breaking it up into smaller pieces which could be individually tested. - Raffi
  • That test was called "MinusCurator", and it sort of overwhelmingly failed. Asheesh began work with Tim on migrating the TripleStoreIndexer to use the new document.add(String, String) method from Nutch rather than LuceneWriter.add(Field, String). The latter is deprecated, and moreover seems to not quite work. Yesterday Asheesh began writing a few helper methods and tests in a branch to help complete this migration.

Next steps

The current goal is to make sure TripleStoreIndexer works. It's pretty deeply broken if we know we can't write a single TripleStore-based value into Lucene. Raffi pointed out that the tag-addition team has a test for this which he thinks already passes.

After that, we will work on landing Tim's and Asheesh's code from the sprint, namely the work on creating new Lucene columns that represent particular RDF predicates, controlled simply by a configuration file.