W3C home > Mailing lists > Public > public-qt-comments@w3.org > August 2002

Data Model and F&O WD: string values of atomic values

From: Jeni Tennison <jeni@jenitennison.com>
Date: Mon, 19 Aug 2002 21:32:41 +0100
Message-ID: <27630187691.20020819213241@jenitennison.com>
To: public-qt-comments@w3.org

Hi,

In Section 5 of the Data Model WD, it says:

  The accessor dm:string-value can be used to recover a lexical
  representation of the atomic value. The details of converting an
  atomic value to its string representation are described in the
  "Casting Functions" section of [XQuery 1.0 and XPath 2.0 Functions
  and Operators]. In particular if the atomic value's type is
  primitive, dm:string-value returns the atomic value's canonical
  lexical representation for that primitive type as specified in
  [XMLSchema Part 2]. If the atomic value's type is derived, the
  lexical representation depends on whether a value is supplied for
  the type's pattern facet: If no such value is supplied
  dm:string-value returns the atomic value's canonical lexical
  representation for the base type (which is a primitive type).
  Otherwise dm:string-value returns a lexical representation that
  matches the value specified for the pattern facet. (This case
  includes xs:integers.)

There are several things that confuse me here, specifically in the
second half of this paragraph.

I can't find the place in the F&O WD where it talks about casting from
a derived type to a string (e.g. from an integer to a string). The F&O
WD talks about casting from a derived type to *its* primitive type in
Section 16.2 and about casting from a primitive type to a string in
Section 16.6, but there's nothing that I can see about casting
directly from a derived type to a string.

There is a hint that you might be able to pipeline the two casts
together in one step, so:

  cast as xs:string(xs:integer('2'))

would be the same as:

  cast as xs:string(
    cast as xs:decimal(xs:integer('2')))

but if you did that then you'd lose information about the fact that
integers can't have any significant decimal places, and you'd get the
string "2.0" rather than "2".

The only mention of the pattern facet that I can find in the "Casting
Functions" section of the F&O WD is where it talks about casting
between two types that are both derived from the same primitive type,
which doesn't apply in this case.

Perhaps when this part of the Data Model WD says:

  The details of converting an atomic value to its string
  representation are described in the "Casting Functions" section of
  [XQuery 1.0 and XPath 2.0 Functions and Operators]. In particular...

It means:

  The normal method of converting an atomic value to its string
  representation is described in the "Casting Functions" section of
  [XQuery 1.0 and XPath 2.0 Functions and Operators]. In this case,
  though...

Or perhaps the F&O WD hasn't yet been updated with the decisions
described in this section of the Data Model WD.
  
BTW, I should note that in the XML Schema for XML Schema, xs:integer
is defined as:

  <xs:simpleType name="integer" id="integer">
    <xs:annotation>
      <xs:documentation
        source="http://www.w3.org/TR/xmlschema-2/#integer"/>
    </xs:annotation>
    <xs:restriction base="xs:decimal">
      <xs:fractionDigits value="0" fixed="true" id="integer.fractionDigits"/>
    </xs:restriction>
  </xs:simpleType>

The pattern facet isn't set in this definition, nor mentioned in the
prose description, although I agree that it should be, since the valid
lexical representations for integers do not include the canonical
lexical representations for decimals. Perhaps you can suggest this as
an erratum.

Cheers,

Jeni
---
Jeni Tennison
http://www.jenitennison.com/
Received on Monday, 19 August 2002 16:32:43 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 16:56:43 UTC