[dsdl-discuss] Potential use cases for complex value validation from ebXML

I am using the term "complex value" to mean anything except simple
tokens or unconstrained strings. It includes lists.

I have been trolling the XML Schema lists for requests that were turned
down for XSD 1.0. I note that Martin Bryan put in a request
UK Data Harmonization Group for something called
"ebXML measure data representation type"

They wanted to be able to validate strings such as
 "10kg" and "5.25lb/ft^2"

Martin: do you still think these are useful/important?
Are they still part of ebXML? (Of course, I am not
asking this because of supporting ebXML per se,
but to support publishing with docuents containing
ebXML fragments or idioms.)

Looking throught the ebXML material,
http://www.ebxml.org/specs/ccCTLG.pdf page 8
they give their type structure as:

Component Types:
 * code type
 * identifier
 * date time
 * amount
 * text
 * measure
Data types:
 * String
 * Boolean
 * Decimal

In http://www.ebxml.org/specs/ebrim2.pdf page 16 is datatypes
 * Boolean
 *String4
 *String8
 *String16
 * longName
 * shortName
 * freeFormText
 * UUID
 * URI
 * Integer
 * DateTime (=xsd:datetime)

Looking through another more helpful spec
http://www.ebxml.org/specs/ccSTRUCT.pdf
and
http://www.ebxml.org/specs/ebBPSS.pdf
their datatypes seem to be

  * text (e.g. rich text?)
  * string
     - including reference or pointer to a document
     - "unique string"
     - UN/ECE Recommendation #21 (???)
     - VAT number or company registration number and national tax identifier
       Can anyone supply examples of these from their locales please?
       In australia it is "[a-zA-z]{3}\s+\d{3}\s+\d{3}\s+\d{3}"
      - token of letters, figures and symbols
      - URL
      --- URL of "a list of codes"
     - MIME type
     - ID/IDREF
     ----UUID within a collection
     - version number (two attributes, major and minor, Integers)
     - postcondition expression: not sure which language is used
     - "specificationElement" and "specificationLocation" not sure why string
     - email address
  * decimal
  * number
    - including base rate expressed as a multiple of 10: 1, 10, 100, etc
  * enumeration of string
  * member of controlled vocabulary
    - ISO 4217 unit of currency
    - UN/ECE UNSPC code list (product codes)
    - ITU E.164 telephone identifier
    - ISO 3166 nation
    - UN/ECE Recommendation # 20 and X12.355 CCT measurement type
        and CCT quality type
    - ISO 4217 currency
    - ISO 639 language
    - ISO 8601 date
  * date and/or time
     - xsd:duration seems common
     - "beginsWhen" and "endsWhen" are represented by strings not
       XSD datatypes: need to check why
  * boolean

and they non-normatively define things like "aggregate"
with reference to UML.

The issue of standard controlled vocabulary is interesting, because if the
controlled vocabulary is dynamic, there is more chance that some schemas
will not be updated and therefore fail spuriously. This suggests the me
that there might be good value in adding some notion of "date of issue"
to controlled vocabularies, so that when validation reports that
there is no such currency as the Euro, the user/developer can immediately see
that it is because of using an old list.

ISO 3166 is causing headache because of the 2 letter and 3 letter forms.
If we can cope with that (two lexical forms, one value space!) we would be
ahead.

It would be great if we could have some way to validate that
IDs are unique within a collection of documents.

ebXML also has some notion of resource bundles for i18n:
the value used is a key to a locale-specific list (i.e. of strings).
This is a kind of one-to-many link that can appear in
multiple values (so is a many-to-one-to-many link). It might
be useful to provide something to support this kind of thing.

Cheers
Rick Jelliffe

--
DSDL members discussion list
To unsubscribe, please send a message with the
command  "unsubscribe" to dsdl-discuss-request@dsdl.org
(mailto:dsdl-discuss-request@dsdl.org?Subject=unsubscribe)