• Pierre Fernique, 2014-05-05
  Comments on tiger team effort and wonders about contents of the 'ivoa' data model.
• Francois Bonnarel, 2014-05-07
  Comments about references to uytpes mappng document.

• Should VO-DMLs Model extend Package?
  Omar notes that this, together with XML schema usage of <sequence>, complicates hand-writing of models.
  ALso, it gives Model a vodml-id element, which can cause problems with the interpretation of that concept and its uniqueness requirements in a model. [Omar: It's more like the other way around. Needing vodml-id for Model is the issue. Also, the fact that the xmls now have imports as their latest element is an annoying consequence, and not only for those "hand-writing" models. I made a couple of suggestions on how to fix this, but I believe that in the end the best approach is to go back to the previous state]

Comments from DM mailing list and private communications

• Mark Cresitello-Dittmar, 2014-04-25
  Comments about the basic 'ivoa' model that is supposed to be used (imported) by all IVOA models

• Pierre Fernique, 2014-05-05
  Comments on tiger team effort and wonders about contents of the 'ivoa' data model.

• Francois Bonnarel, 2014-05-07
  Comments about references to uytpes mappng document.
• private communication from Kristin Riebe: why is there no counterpart to UMLs aggregation relation?

Open issues identified by editors/authors

• Should VO-DMLs Model extend Package?
  Omar notes that this, together with XML schema usage of <sequence>, complicates hand-writing of models.
  ALso, it gives Model a vodml-id element, which can cause problems with the interpretation of that concept and its uniqueness requirements in a model.

Comments from DM mailing list

• Mark Cresitello-Dittmar, 2014-04-25
  Cmments about the basic 'ivoa' model that is supposed to be used (imported) by all IVOA models

Open issues identified by editors/authors

• Model inherits from Package. This, with use of <sequence> in XSD causes some model specific features such as authors, version etc to end up at the end of a Model definition. Is this cumbersome? Reasons we did this is to inherit common elements like vodml-id, name,d escription and all the type collections. Also, UML has the same patterns, i.e. Model is-a special kind of Package there.
• vodml-id-s should be unique in document for all elements not a <model>. Should we extend this rule to <model> as well? Note that when generating vodml-id using generaiton rules a package with the same name as the vodml-id of the model would end up with same vodml-id, hence no uniqueness. Ofcourse one need not use such rules, or can give package a vodml-id different from its name to resolve this. Alternatively we could insist on using model's vodml-id as prefix for vodml-ids as well. Currently only utypes must have a prefix.
• When referreing to a vodml-id using a utype element one always MUST use as prefix the vodml-id of the model, whether an imported model or the model itself. It was a decision of the tiger team to make such prefixes "globally unique", i.e. every usage within the IVOA MUST use the same prefix, which MUST be the vodml-id of the model. This in contrast e.g. to XSD namespace/xmlns where prefixes can be freely declared. Sould we revisit this?
• ModelProxy has an element named prefix, which currently MUST be the same as the vodml-id of the imported model it represents. Is this name, prefix, ok?
• ...

• Arnold Rots, 2014-04-24:
  • Rename ReferencableElement to RefenceableElement.

• Mark Cresitello-Dittmar, 2014-04-25
  Cmments about the basic 'ivoa' model that is supposed to be used (imported) by all IVOA models
  • Add TimeValue to ivoa model.
  • Add ivoa:URL as sub-type of ivoa:anyURI.