Re: [xsl] Schema-awareness clarification

Subject: Re: [xsl] Schema-awareness clarification
From: David Carlisle <davidc@xxxxxxxxx>
Date: Wed, 28 Feb 2007 11:46:37 GMT
> 1)
yes (I think)

> 2)
> Schema-awareness means XML Schema schema awareness, period.

Not necessarily. Just as a non schema aware (or XSLt1 system) can (and
some do) accept data that isn't in fact parsed by an XML parser, but is
just a constructed view that has the same features (such as a DOM
generated programmaticlly or a pasrser for a non-xml language that
generates SAX events) So a schema aware system can take a type annotated
input tree from anywhere, the most natural place is from a PSVI
generated by a W3C schema validator, but the XSLT spec doesn't mandate
that this is wher it has to come from.

> Even when other schema languages like RELAX NG make use of XML
> Schema's data types, such schemas can not be imported into XSLT using
> the import-schema element. This element's true name is "import an XML
> Schema schema".

This part is essentially true though.

>3)
>When we talk about the input document nothing disallows an XSLT
>processor, XSLT 2.0 or XSLT 1.0, schema-aware or not, to have some
>extension option validating such input file. 

You can use a validating parser (dtd/xsd/relax/...) on the input but the
basic conformance level explictly forbids a system from accepting type
annotated trees, so you can not use any types infered from this
validation if working at this level.  It would be possible for a schema
aware system to also have an extension function that allows validating
against a relax schema that uses a type library, but a system that
-only_ supported such an extension and not XSD schema wouldn't be
meeting either of the two specified conformance levels for XSLT2.

>4)
>Likewise nothing disallows an XSLT processor to have some extension
>feature that can validate an output files against any schema language
>supported in this sense. But such a feature has nothing to do with
>"schema-awareness" in the XSLT 2.0 sense, and is so to speak outside
>the XSLT spec.

True.

David

Note that I'm _not_ on the working group, this is just my reading of the
spec....

Current Thread