W3C home > Mailing lists > Public > www-xml-query-comments@w3.org > January 2002

Data Model WD - canonical lexical representations

From: Jeni Tennison <jeni@jenitennison.com>
Date: Sun, 13 Jan 2002 14:21:37 +0000
Message-ID: <58539046948.20020113142137@jenitennison.com>
To: www-xml-query-comments@w3.org

Looking at the string-value accessor for simple typed values - while I
think it's right to use the canonical lexical representation of the
values, unfortunately the XML Schema Datatypes Rec doesn't detail the
canonical lexical representation of all of the primitive types. In
particular, no canonical lexical representation is specified for:

  - xs:string, xs:base64Binary, xs:anyURI (but that's OK, I think we
    can guess)

  - xs:duration - presumably the lexical representation contains all
    components of the duration (years, months, days, hours, minutes
    and seconds, even those that occur 0 times? Or are these omitted?
    In the latter case, what's the canonical lexical representation of
    PT0S? Since the number of seconds can be a decimal, is this
    decimal represented with a decimal point (i.e. using the canonical
    lexical representation for xs:decimal)?

  - xs:date - what happens to the timezone component? Presumably,
    unlike xs:dateTime and xs:time, this isn't normalized to Z?
    (And similarly for xs:gYearMonth, xs:gYear, xs:gMonthDay,
    xs:Month, and xs:Day)

  - xs:QName and xs:NOTATION - these are the trickiest (their value
    spaces are the same). The XML Schema Rec states that the lexical
    representation of a QName depends on the in-scope namespaces. When
    you're doing a query/transformation, which namespace declarations
    do you use - the ones in the query/stylesheet or the ones from the
    source document? What if there's more than one namespace
    declaration for the namespace URI? What if there aren't any?

    The difficulty with using the original prefix for the QName is
    that the definition of a canonical lexical representation is that
    two equal values have exactly the same canonical lexical
    representation. Equality of QNames should be based purely on the
    namespace URI and local name, not on the prefix. One possibility
    would be to introduce something like:


    but this is not, of course, a valid lexical representation of a


Jeni Tennison
Received on Sunday, 13 January 2002 09:22:36 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:21:15 UTC