OEWG 20110830

From GO Wiki
Jump to: navigation, search

OBO-Edit Working Group Meeting: Tuesday, August 30, 2011, 8:30am PDT

Conference call numbers:
US: 1 866 953 9688
UK: 0808 238 6001
PIN: 801561

Agenda/Chair: Nomi Harris
Minutes: Nomi
Attendees: Nomi, Heiko, Midori, Rebecca, Ramona

Bug and Feature Trackers

Fixed since last meeting

  • Fixing JUnit tests for OBO and OBO-Edit
    • Fixed some tests; removed some that Chris M. said were obsolete
    • 2.1-b18 now passes a lot more tests than 2.0
    • 2.0:
      • 21/43 OBO JUnit tests pass
      • 1/24 OBO-Edit JUnit tests pass
    • 2.1-b18:
      • 38/38 OBO JUnit tests pass
      • 11/24 OBO-Edit JUnit tests pass
        • None of the failures are new--all 14 that fail now also failed in 2.0.
        • Trying to fix at least the obo2obo and obodiff tests

Discussion Items

  • New file browser had bug in b17--Heiko quickly fixed it and we released b18 the next day.
  • Did anyone (other than Midori) get a chance to test obo2obo to make sure my fix for [ geneontology-Bugs-3316815 ] (filtered save doesn't save dangling parents) didn't break anything in obo2obo?
    • Midori found one issue in obo2obo but it turned out to be old and unimportant
    • Fixed problem where obo2obo recently stopped printing status messages to stdout.
    • The obo2obo tests I tried yielded the same output for OE 2.0 and 2.1-b18.
  • Amelia asked what the OE policy on repeated words was, because she was getting the warning for strings like "bilirubin-glucuronoside = bilirubin + bilirubin-bisglucuronoside".
    • The policy for repeated words is that you get a warning if there are repeated words unless they are on the list of allowed repeated words (which you can add to yourself) or--theoretically--they are in different sentences (e.g., "Protein plays a role in eye development. Development starts etc"). I say "theoretically" because this is not working right--you still get a warning if you have a period between the repeated words. This is because the imported method for checking whether the second word begins a new sentence (com.swabunga.spell.event.StringWordTokenizer.isNewSentence) doesn't work--it returns false even when the second word does start a new sentence (e.g., "development. Development").
      • Symbols such as + and = don't exempt repeated words from the repeated word warning. It would be possible to add this (probably a few hours of work) if there is agreement on the list of symbols that should exempt repeated words from the warning. (I don't think it would be possible to make "+" and "=" count as words with the current setup, because the tokenizer doesn't treat symbols as words. I'm not sure you'd want that to be the case generally, anyway.)
    • Suggestion: use the Configure buttons in the Verification Manager to turn off tests (such as repeated word or repeated whitespace) that you find annoying rather than informative.
      • After agreement from at least one OEWG member, Nomi set the repeated whitespace test (which complained about two spaces between sentences, for example) to default to "Never" (users can turn it on with the Configure interface).
    • See related bug report verification checks--repeated words, sentence boundaries, which requests (among other things) that repeated words not count as repeated if there's punctuation between them.
  • Next meeting: we will skip Tuesday, September 6 and meet next on September 20.
  • We will then move to a once-a-month meeting on the first Tuesday of the month (Oct 4, etc.)

Currently working on

  • Working towards making an official 2.1 release
    • Mike Cherry wrote, "For the production site I'm still using OBO-EDIT 2.1-beta1. The OBO2OBO script is run nightly to convert an OBO v1.0 file from the OBO v1.2 and to make an OBO diff file. If we should be using a newer version of OBO-EDIT please let me know which one is best. The need on the production site is limited to obo2obo and obodiff. To determine if the script failed I consult the returned status and I don't use the output to STDERR."
  • Extended Characters not being caught
    • There's a Text Editing configuration option (in the Configuration Manager) "Allow extended characters". In 2.00, OBO-Edit complained about extended characters (e.g., in "Perú") whether or not this option was checked. Then in one of the early versions of 2.1, this reversed: Perú was silently accepted regardless of the setting of "Allow extended characters". So in both old and new OE, the "Allow extended characters" option seems to be ignored.
    • Proposal:
      • Remove (comment out) the non-functional "Allow extended characters" configuration option. I added a bug report about this.
      • Continue to allow diacritical markings (e.g., "Perú") in the definition, comments, and synonyms (i.e., don't change current behavior).
      • Continue to complain about these in xrefs (i.e., don't change current behavior).

Next bugs to work on

  • Fixing any bugs that the JUnit tests reveal
  • Is_a closure broken
    • Midori commented, "I can confirm that I get the same thing (or at least something similar) happening with 'regulates' relationships: is_a closure saves some terms that would only be there if it followed at least one 'regulates' link.
      • I am looking for clarification on this. In Midori's example, it looked to me like it didn't have to follow "regulates" links to save the terms in questions, but I may be missing something.
      • Midori is going to look at this again, and Nomi will look for other is_a closure test cases and rerun them.
    • Chris said "This has become a lower priority for OE, since we will be able to do most of this in Oort."
      • Heiko will talk to Chris about sending out a message about Oort

Discussion

Links

Back to OEWG meeting agenda and minutes list