OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help



   Re: [xml-dev] implications of xpath 2.0 for schematron, XSD

[ Lists Home | Date Index | Thread Index ]

Schematron is expected to become an ISO standard in about 3 months now, 
barring unforeseens.
I expect it will gently become part of the fabric over the year after 
that. I don't anticipate standards
bodies will put out their own Schematron schemas particularly, more that 
they will point to
Schematron as a good way for users to test their additional constraints.

Bryan Rasmussen wrote:

>I was wondering if anyone has given thought to the 
>implications of xpath 2.0 for schematron (possible future xpath 2.0 using
>schematron), especially as it relates to the PSVI. 
I have! There is a little implementation change for xpath 2, relating to 
treating empty
results as boolean false, but there is no reason that Schematron cannot 
use Xpath 2.0
operating on the PSVI. In the jargon of the standard, that is just a 
different "query language

>As lots of people have been using Schematron as an extension to other schema
>languages, via the use of embedded Schematron, what would the implications in
>that situation be? 
The ISO DSDL framework does not have anything in a particular about 
Schematron embedded
in XML Schemas (or in RELAX NG for that matter), ISO Schematron allows 
it. There is an
annex F Using Schematron as a vocabulary that blesses the practise. See

It is just a matter for the implementation to validate the document with 
XML schemas before
validating with Schematron.

>In one way I think it would be great, in that one thing I have found Schematron
>useful for is developing algorithmic checks on text content, and XPath 2.0
>extends the ability to do this greatly. 
Yes. I am not sure that XPath 2 gives that much more than EXSLT supports 
and I am not convinced that the PSVI provides anything that casting 
allows in expressions.
But technical power is less important than familiarity: as more people 
get more savvy
at XPath 2 over the PSVI, it will be natural for them to want to add 
constraints, and
ISO Schematron is fine by that.

>Finally, one of the things that have often worried me vis-a-vis the future of
>Schematron is that I figured at some point it would just be absorbed into W3C
>Schema, given that it was just too useful not to be. But the concept of W3C 2.5
>or something, with Xpath capabilities and a reliance on Xpath 2.0 seems uhm
>scary. Of course that could conceivably also be a problem in the context of
>xsd:unique, though obviously not as hairy. 
We tried to make ISO Schematron W3C-friendly (well, friendly to 
every-one) by allowing
elements to be used in isolation.  So if they had a demand for 
constraints, and wanted to add
sch:assert (or, better, sch:rule or sch:pattern) as a standard extension 
element, there is nothing
stopping them from the ISO standard side.

xsd:unique and xsd:key are currently not typed, but they are streamable 
in XML Schemas.
Many people made noises that there could have been a better system, but 
I haven't seen
much in the way of concrete suggestions.  It would be interesting to 
know how many
implementations of XML Schemas for data-binding or application 
generation actually
use the unique and key declarations to do something special. Otherwise, 
a general purpose
constraint language is just as (well, a lot more) powerful.

At one stage MSM sounded me out about making up a W3C Technical Note about
using Schematron with XML Schemas or something (probably I would have nicked
Eddie Robertsson's material) but it never eventuated. W3C schema people 
have always
been quite encouraging about Schematron; they see it fits well as a 
constraint layer.

Supporting Schematron would be an easy option for them, and probably 
would reduce
distractions from their efforts. The main things working against it are, 
I would say,
that they are so busy with XML Schemas and XQuery they don't have the 
or it is outside their product lines, or that XML Schemas operates 
streaming and would be
a better fit with a steaming version of XPath, such as STX, rather than 
full random-access
XPath (I don't think this is a problem, myself.)

B.t.w., I am presenting a day seminar on Schematron in Beligum next 
week.  Schematron
will also feature in QA seminar in the same series. For more info

Rick Jelliffe


News | XML in Industry | Calendar | XML Registry
Marketplace | Resources | MyXML.org | Sponsors | Privacy Statement

Copyright 2001 XML.org. This site is hosted by OASIS