(This comes from a suggestion to make an explicit pattern for IDREFS in
the RNC schema for ISO Schematron, to make validation of incomplete
schemas easier.)
G. Ken Holman wrote:
> Thanks, David, for this approach ... it had not occurred to me.
>
> Note, though, that while the RNC schema is informative, the RNG schema
> is normative. I'll admit that while the *expression* changes with
> your suggestion, the validation semantics have not changed ... so this
> should be innocuous ... all you are doing is making the expression of
> the semantics flexible to be overridden.
I don't have any problem with making this change.
Also, I don't think it matters that the RNG schema is normative, because
it normative for validating a schema. What Ken is wanting to do is
validate an *incomplete* schema; it is fine to have a version of the
schema that does this (with appropriate documentation.) Actually, this
is the kind of thing that I think validators should provide options for
checking: akin to the "feasible" validation that Toplogi and Jing
provides, though I don't think that Jing goes as far a not checking
references.
At the analytical level, a schema expresses the constraints of a
complete document. But an application should be free to use (or not use)
this information in any way when a document is being constructed. The
aim is to have as much validation as possible as early as possible in a
pipeline, but not too much or too early! Now Schematron formalizes this
into phases, but for other schema languages it is an
implementation-dependent matter.
Could it be possible to make a phase language for RELAX NG? Borrowing
the Schematron syntax, you could treat validating each XSD datatype as a
different pattern, and also validation of particles by RELAX, and turn
them on or off.
<phase:phase id="all">
<phase:active name="xs:ID" />
<phase:active name="xs:IDREF" />
<phase:active name="rlx:attributes" />
<phase:active name="rlx:text" />
</phase:phase>
This is something that could presumably go into ISO DSDL as part of part 10.
Cheers
Rick
-- 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)Received on Wed Apr 4 11:17:20 2007
This archive was generated by hypermail 2.1.8 : Wed Apr 04 2007 - 11:38:03 UTC