Minutes of Atlantic UBL TC call 14 April 2005

[jon.bosak@sun.com]

MINUTES OF ATLANTIC UBL TC MEETING
15H00 - 17H00 UTC WEDNESDAY 14 APRIL 2005

ATTENDANCE

   Jon Bosak (chair)
   Mavis Cournane
   Mark Crawford (vice chair)
   Stephen Green
   Mike Grimley
   Anne Hendry
   David Kruppke
   Ken Sall
   Paul Thorpe

STANDING ITEMS

   Additions to the calendar (http://ibiblio.org/bosak/ubl/calendar.htm)

None.

   Liaison reports

None.

   Subcommittee reports

SSC not meeting this week while ver team discussions are underway.

   Team report: Versioning

      StephenG: The team met for an hour Tuesday.  We discussed
      three possible ways forward, two of which were considered
      viable.  The nonviable alternative is to not use schemas but
      rather some other mechanism.  The first viable alternative
      is to redeclare everything and just issue a new set of
      schemas; this is the default if no better method is found.
      The second viable alternative is to preserve polymorphism
      through a qualified application of substitution groups that
      does not use abstract types and thus avoids the usual
      problems associated with substitution groups; this approach
      will be worked on over the next few weeks.  ArofanG will
      produce a fragment showing the proposed method by this
      weekend, and StephenG will produce a prototype.  Either
      option will require changes to the NDRs.  The participants
      so far are DavidK, StephenG, ArofanG, and TonyC.

      Discussion followed.  Highlights:

       - In practice, "polymorphism" means that a 1.0 application
         should be able to process a 1.1 instance if given 1.1
         schemas, so that people don't have to change all their
         software just to cope with a minor version release.  It
         also means that we could use XSD to prevent incompatible
         extensions or restrictions rather than imposing the rules
         ourselves.

       - Without "polymorphism", backward compatibility would
         still exist at the data level but not the application
         level.

       - The purely local approach adopted by ATG2 means that no
         solution based on substitution groups will work for them.

       - The design goals and use cases for this discussion were
         covered in the ModNamVer paper.

       - If we redeclare everything and say that polymorphism is a
         goal for a later release, we could still reasonably call
         the upcoming release 1.1.

       - There may be a case for aligning with ATG CC modules in
         1.1; to do so, we would have to drop the goal of complete
         backward compatibility with 1.0 instances.

      StephenG: The ver team meeting scheduled for Friday has been
      canceled, as most of the work is being done by email.

      AGREED: The ver team should first work through schema and
      instance examples, and then follow with candidate NDR rules.

ACTION ITEM REVIEW

   ACTION: JonB to send Hangzhou invitation letter and logistical
   information to the whole ubl list.

      Status: Done.

   ACTION: MikkelB to send DKLSC info (co-chairs, etc.) to JonB;
   JonB to propose DKLSC formation to the TC.

      Status: Overdue.

   ACTION: JonB to resume the discussion with Altova.

      Status: Pending release of SBS.

   ACTION: MartyB and TonyC to create a briefing package
   summarizing the issues for the TC and send it out the week of
   25 April so that we can prepare for the discussion.

      Status: Pending.

   ACTION: JonB to arrange a meeting to discuss the versioning
   problem.

      Status: Done.

NDR WORK SESSION

   MavisC took over for the NDR team.  Discussion began on the
   list of NDR issues, but it didn't get very far away from the
   versioning question.

   Question: What does "minor version" mean?

   MarkC: It means that we only do optional extensions, optional
   refinements to facets, optional attributes, optional
   classes... This is what's in the NDRs as they now stand.  So if
   we can't figure out how to implement polymorphism, we can just
   drop the NDRs that relate to it.  A major version is one that
   has restrictions or new mandatory elements or that eliminates
   mandatory elements.

   JonB: Let's pick this up next week after we see the outcome of
   the ver team work.

Jon Bosak
Chair, UBL TC