OEWG 20110419

From GO Wiki
Jump to navigation Jump to search

OBO-Edit Working Group Meeting: Tuesday, April 19, 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 Dietze, Ramona Walls, Paola Roncaglia, Harold Drabkin, Midori Harris, Karen Christie, Thomas Waechter

Bug and Feature Trackers

Discussion items

Fixed since last meeting

  • Configuration manager not opening (on some Linux machines)
    • Was throwing an exception while trying to read from oboedit.vmoptions, which isn't there on Laurel's Linux box
    • Still not sure why vmoptions isn't there (though it doesn't seem to hurt OE), but now it doesn't throw an exception, so Config manager comes up.
  • The string "et al.," causes warning about whitespace
    • fly_anatomy_XP.obo has comments and definitions that include the phrase "et al." followed by a comma. OE was, annoyingly, complaining about those.
    • I added a special case to exempt "et al." from this complaint.

Currently working on

  • Cycle over mixed relationship types -> infinite loop
    • Can't save fly_anatomy_XP.obo because OE goes into an infinite loop while running verification checks.
    • Chris thought this was because fly_anatomy_XP has cycles over mixed relationship types, which is valid but was not considered valid when OBO-Edit was originally developed. This may be true, but in fact it turned out that the infinite loop was caused by a term that listed itself as an alternate ID. The validation process involves recursively checking the term's alternate ID, and since it's the same term, it just recurses over and over.
    • Two ways to fix the infinite loop (could do either or both):
      • Before recursing on alt_id in the validation check, check whether it's the same as the primary ID.
      • When reading in ontology, check for alt_id == primary_id and refuse to add it.
  • Is_a closure broken
    • Filtered save with "is_a closure" checked doesn't save all the terms it should.
    • Fix in progress
  • Search for children misses some (unless reasoner on)
    • If you turn on one of the reasoners, you find more children. Maybe the "child" and "parent" search aspects should be disabled if the reasoner is off?

Next bugs to work on

Discussion

  • Introducing Heiko Dietze. Heiko will mostly be working on a next-generation web-based ontology editing tool, but will be looking at OBO-Edit for inspiration, and may occasionally help fix OBO-Edit bugs.
  • DirectLinkReasoner (Chris's new simple reasoner)
    • Ramona tested and found some bugs.
    • Probably need to leave DirectLinkReasoner out of next release until more debugging and testing can be done.
  • Paola sometimes has to force-quite OBO-Edit (on her Mac). No one else has had this happen. Paola will email Nomi her oboedit log file the next time this happens.
    • Paola said it hasn't happened this week.
    • Harold said this has occasionally happened to him in the past.
  • Should we disable "child" and "parent" search aspects if the reasoner is off? Fixing those searches might be hard.
    • Still need to discuss with Chris
    • Harold said that in GO, links are asserted, so you should be able to find all the children without needing the reasoner
    • Midori said that, nonetheless, OE doesn't find all the children it should in GO.
  • Bogus warning messages during validation: if you have a URL in a comment or definition, you get complaints about there not being spaces between "sentences" and lack of capital letters. Not sure what's the best way to address this. URL is not parsed as a single "word" but rather as multiple "sentences".
  • In fly_anatomy_XP, the verification check goes into an infinite loop because there's a term that has an alternate ID identical to its primary ID.
    • The consensus was that you shouldn't ever have alternate ID == primary ID, and OE should pop up a fatal warning when this happens, just as it does with many other illegal constructs.
    • Nomi will check whether this is already happening if an obsolete term has a replaced_by or consider ID that is the same as its primary ID--this is also illegal.
  • Bug causes Text Editor in autocommit mode to lose your changes if you add a word to dictionary
    • This is probably due to Nomi's recent change that made words turn black immediately when they were added to dictionary. However, she apparently didn't test this in autocommit mode.
    • Ramona always uses autocommit mode, and wants to know why would anyone use manual commit mode?
    • Harold said that's what he's used to, and he also likes the idea that if he leaves a term and has forgotten to commit, it warns him, so he realizes he's not actually done editing it. Also, what if he accidentally edits something that he didn't mean to? Autocommit would just silently commit the unwanted change and he wouldn't notice.
  • Warning about ID profile mismatch: need smaller example
    • Midori offered to add ID profiles to the tiny test ontologies so that Nomi can test and debug this.

Links

Back to OEWG meeting agenda and minutes list