[dsdl-discuss] DSDL Part 5 (DTLL)

ISO DTTL is based on Jeni Tennison's ideas for defining data type
libraries using properties. DTTL would be a good companion to Schematron:
  * use standard type libraries developed for RELAX NG or XSD
compatability for free
  * reduce the number of home-made assertions
  * implementable on top of XSLT 2, so may be low-hanging fruit

I am wondering on the best way to go about it:

1) Extension function that can validate a type

<sch:rule context="eg">
  <sch:assert test="dt:validate(@date, 'types:japanese_date')" >
     The date attribute should be a Japanese date.
  </sch:assert>
</sch:rule>

Pro: fits into XSLT extension function mechanism. Just exended query
language binding
Con: fails on XSLT 1 systems

2) Foreign attribute

 <sch:rule context="eg">
  <sch:assert test="true()" subject="@date" dt:type="types:japanese_date" >
     The date attribute should be a Japanese date.
  </sch:assert>
</sch:rule>

Pro: doesnt fail on XSLT 1 implementations, declarative, exploits
@subject, clean,
  fits in with the idea that we are not assigning type to the
context--the context is just
 a convenient pivot
Con: test="true()" looks funny

3) Foreign element

 <sch:rule context="eg/@date">
  <sch:assert test="true()"><!--satisfy content model--></sch:assert>
  <dt:datatype type="types:japanese_date" />
</sch:rule>

 <sch:rule context="eg/@date">
  <sch:assert test="true()"><!--satisfy content model--></sch:assert>
  <dt:datatype library="xtypes.dtl" type="types:japanese_date" />
</sch:rule>

Pro: clean, doesn't fail on XSLT 1 implementations
Con: extra assert to satify content model is ugly, looks like a type
assignment to the context which is not the way to go

I favour 2) so far. Any ideas?

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)