[dsdl-discuss] Re: First draft of Schematron spec (Schematronin one defun)

From: "Eric van der Vlist" <vdv@dyomedea.com>

> Some comments about this draft:
>
> * I wonder if this is wise to include the list of the parts of
> DSDL in the foreword: the list you've provided is out of date
> and may change again in the future...

That comes from the stylesheet. ISO rules don't allow reference to
unpublished parts, I believe.

> * I like the idea of defining language bindings.

That is needed to
  1) future proof for XSLT 2, XPath2 and XQuery coming online
  2) reflect actual practise (Schemarama uses Squish query language,
    someone used XPath not XSLT, someone used EXSLT)
  3) simplify the spec: it makes it easy to delegate all mention of
    XSLT to an annex, which is nice, and I think will make it easier
   for any implementers
  4) present Schematron in a way disconnected from XSLT: the typical
   way people see Schematron is a hack to simplify validation in XSLT.
   While I exult in that, Schematron's abstract patterns are part of
   a much longer journey in trying to describe things that go on in documents.
   I have almost finished the example schemas for a non-normative annex
   which gives a pattern library ("mixed-content, list, name-value-pair, query,
   head-body, titled-container, etc." )

> * Making XSLT the default binding provides backward compatibility
> with the current version of Schematron.

Yes.

> * I am not sure whether it is enough to keep the same namespace
> since you have introduced new features which will not be
> supported by existing implementations of Schematron: "new"
> implementations will have no problem with existing schemas, but
> the opposite will be more problematic.

There has been a version number in effect for Schematron since version 1.5.
Schematron implementations that don't implement version checking (and I
think none do, but it is trivial to add) are not implementing the spec, to
some extent. Blame the victim!

Until ISO schematron, schematron schemas have not been really exchangeable
in any case: figuring out the best place for <key> has been enough for that,
plus the different query languages. This is not a sad thing: in fact, I have tried
to encourage diversity and experimentation to shake out the design a bit.

Now that the design has done being shooked ( schema/rule, assert/value-of )
it is time to standardize: I am trying to make the spec so that the work implementers
need to do to support the standard is very small. But now that we are
standardizing the focus of course changes from encouraging variants
to allowing well-described variants but discouraging non-standard implementations.

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)