[dsdl-comment] Re: XML Data Design to Facilitate Compound Documents

> Surely it's not inherent in the document whether it's validated using
> a single process or using multiple processes.

But it might be inherent in the document whether it is validatable in a
single process. If the document has no non-regular constraints, then it
inherent that it can be validated in a single process (by a conventional
regular grammar-based system.)

Then you might say that a document which has both regular constraints and
non-regular constraints is inherently a multiple schema-language document,
given the current range of schema languages.

But I don't think Murata-san wants the term because it represent some
sciency property of the document, but as a practical matter on whether it
has (or must have) multiple schema modules, given the state of the art. A
software engineering category, not a theoretical computer science
category.

I think that whether there is one process or multiple processes is not so
much an issue as whether the processes can be effectively parallelized to
fit into a concurrent streaming implementation on the one hand, and the
issue of whether there needs to be multiple files and setups on the other.

Even that is not a very sharp distinction now, given that now we can
convert most XSDs (a converter I wrote for JSTOR) and most RELAX NG
schemas (via XSD in Jing at the moment) to Schematron and thence XSLT, so
we have an existence 'proof' that DSDL can be validated with some unified
software rather than by unrelated processes (I don't know about NVDL and
XSLT). And given that (all?) XPaths can be rewritten into forms that only
look on the forward axis (even if they must be teased apart to do so) it
is possible to have a one-pass Schematron implementation too (well, a
match-reduce implementation) though it may not be efficient. So the issue
of the number of processes looks like it is receding into an issue of
implementation strategy rather than a hard practical constraint.

Cheers
Rick

--
DSDL comments
To unsubscribe, please send a message with the
command  "unsubscribe" to dsdl-comment-request@dsdl.org
(mailto:dsdl-comment-request@dsdl.org?Subject=unsubscribe)