Minutes of UBL TC meeting 17 August 2006 (1)

[jon.bosak@sun.com]

UBL TC MINUTES FOR THURSDAY MORNING 17 AUGUST 2006

Members attending the Montral meeting in person:

   Jon Bosak (chair)
   Mikkel Brun
   Mavis Cournane
   Mike Grimley
   G. Ken Holman
   Andy Schoka
   Paul Thorpe

Observer attending in person:

   Christian Lanng

Members attending by phone:

   Tim McGrath (vice chair)
   Peter Borresen

   The phone line was open starting around 9 a.m.

Support package scheduling (carried over from Monday)

   "UML and EF data models (pointers) (McGrath)"

      ACTION: TM to upload the Enterprise Architect UML data
      model; JB to link this from the UBL 2.0 index file and add a
      subsection for this in the same section as the document
      models.

      TM: Is the EF model in sync with the final schemas,
      including the changes that GKH handcrafted?

      MC: We have confirmed with SW that these changes have been
      incorporated in FX; see posting to the list.

      [At this point PB joined the call, and in the subsequent
      discussion of the use cases, we forgot to assign an action
      item covering publication of the EF data model.]

   "Danish use cases and example instances (Borresen)"
   "Sample business rules expressed in schematron (Borresen)"
   "Input Xforms (Borresen)"

      PB: We will deliver a package containing sample documents,
      scenario descriptions, schematron and XSL transformations,
      and XForms when we give our presentation at the November TC
      meeting in Singapore.

      AGREED that these materials should go into the Support
      Package as "Sample Implementation Guide from Denmark."

   "Output stylesheets (maybe by reference) (Holman)"

      GKH: HISC still dormant due to lack of resources; will not
      have stylesheets done by November, but will have one for
      Invoice demonstrating stylesheet generation.

Proposed customization methodology

   GKH walked us through his proposal on "UBL 2.0 subsets,
   extensions, versions, validation and interchange":

      http://www.oasis-open.org/committees/document.php?document_id=18849

   The xpath-based methodology developed for the UBL SBS turns out
   to have a lot of uses.  The xpath file enumerates all xpath
   addresses of all information items in a given document type,
   thus allowing programs to see the hierarchy of elements;
   schemas only show parent-child relationships, but xpath files
   show ancestor descent relationships as well.  Given certain
   other conditions (e.g., use of global types throughout), this
   capability allows xpath files to be used to programatically
   verify that a given schema defines a proper subset of another
   schema -- most importantly for us, to guarantee that any
   instance validating against a particular UBL subset schema will
   also validate against the standard UBL schema for that document
   type.  The proposal uses this capability to provide a set of
   tools for verifying conformant UBL subsets.  The tools will be
   made available from the Crane Softwrights site under some form
   of non-assertion covenant.  Note that the methodology says
   nothing about how a subset is created; it just provides an
   automatable way to verify it for compliance.

   Benefits of this approach include:

    - Automates verification of subset compliance with standard
      UBL schemas

    - Allows the generation of XSLT files that can be used in the
      second UBL validation phase to automatically strip out info
      items defined by minor UBL versions later than the version
      that a given implementation will support; this means, for
      example, that the NES group can publish an "NES Subset
      Filter" that will turn any UBL instance into an NES instance

    - Provides a formalization for a shared "validation shell"
      that can be specified in trading partner agreements.  In
      support of this capability, GKH provides an enhanced version
      of Schematron 1.5 that implements the ISO Schematron
      "include" directive

    - Removes any dependence on problematic XSD
      extension/restriction mechanisms for doing customizations

    - Can apply also to minor versioning (because the major
      version is a subset of the later minor version)

   Since this approach caters to the user need to have everything
   possible defined using XSD schemas, UBL SBS 2 will be defined
   using a set of actual subset schemas rather than just a set of
   xpath files.

   AGREED that "customized schema" is probably a better term than
   "subset schema" because the customized schema can in fact
   contain extensions.

   AGREED that the proposed methodology is a significant
   contribution and should be written up as our customization
   methodology, to be published as a Committee Specification.

   AGREED that the proposed methodology does not obviate the need
   for a separate, later CS on customization guidelines, i.e.,
   guidelines for how to actually create customizations.

   AGREED that the methodology points toward a possible strategy
   for minor versioning that we will discuss further after the
   release of UBL 2.0.

Third public review

   AGREED that we will perform a third public review to fix a
   couple of problems discovered in the second public review.

   AGREED to allow the PSC to determine the specific actions to be
   taken to effect the necessary repairs.

   AGREED to include a number of other changes previously
   identified as candidates for revision in UBL 2.1.

   AGREED to accept the offer of the Danish team to manage
   production of the revised schemas.

   AGREED on the following workflow:

    - JB produces a detailed schedule of this review cycle ASAP

    - UBL PSC decides the disposition of issues 11 and 15

    - As soon as the PSC has disposed of 11 and 15, TM publishes
      the PRD3 change log (including the implementation of a
      number of other changes that would otherwise have been
      deferred till 2.1)

    - PB changes the spreadsheets per the change log and generates
      a new set of schemas

    - In parallel, MG manually revises the PRD2 schemas per the
      change log to produce a parallel set of schemas

    - PT converts both sets of schemas to ASN.1 and performs a
      diff for Q/A purposes

    - When we're satisfied that the schemas have been correctly
      revised, we put the resulting PRD3 to a vote and move to
      public review

   AGREED that we will bundle formal approval of the changes into
   the vote on the PRD3 Committee Draft.

Jon Bosak
Chair, OASIS UBL TC