Lists Home |
Date Index |
From: "Doug Ransom" <Doug.Ransom@pwrm.com>
> Does anyone out there use schematron in addition to schema? I am guessing
> schematron has about 0.1 % of market share. It looks like an interesting
> technology we may use to define some of our document formats that cannot be
> completely described in schema.
I am happy to report that, there being no Schematron implementations that cost money
(or which don't come bundled with some other system), Schematron has 0% market
share :-) And we are hoping to double that in the next month!
More seriously, the ASCC Schematron website gets over 80,000 hits a year.
Topologi's registration list (for the free Windows Schematron/XML Schemas
validator front-end) shows a very widespread interest: I had expected more
downloads from academics or universities, but it is almost entirely commercial
companies from what I can see. And we seem to get just as much interest
from Europe as North America, perhaps because Schematron does not
need localization in its validation messages (i.e., you write a German
schema, you get German messages.)
There are now a few more books published with chapters on Schematron
(Francis Norton's, and the recent WROX books) so there more awareness that
this thing exists now. I have always thought that Schematron (and XLinkit)
could only become popular after XML Schemas had been released for a year,
because people's minds would be occupied with figuring out that.
After people start to figure out whether they need XML Schemas (which
probably relates to how much you need data typing) rather than DTDs
(or RELAX NG), Schematron's strengths (and weaknesses) will be
easier to understand. There have been some good first steps in
evaluating Schema languages: the XML 2001 conference had
an interesting talk from Eric van Vlist (available online, also
paraphrased in an XML.COM article) and also had a very
in depth Schema panel: I hope the panel judges will be publishing
their comments soon.
Schematron is being developed into an ISO standard currently. There have
been very few suggestions for augmentations or deletions over the
last two years. I am not sure what the best thing to do about
XPath 2 and ISO Schematron. Probably we would just take
XPath2 and disallow any PSVI bits.
I invite anyone interested in discussing Schematron evolution
(or other rule-based schema languages) to join the Schematron-love-in
> If path or semantics dependancies exists in an XML document, I think it is
> reasonable to expect clients to validate with schematron if they choose;
> otherwise, they can use schema with some custom imperative code (i.e. VB) to
> verify the semantic dependancies.
Yes, it is good for many practical purposes. I still am positioning Schematron
as an addition to other schema languages (DTDs, RELAX, XML Schemas)
because the grammar-based languages are supported by the big vendors
(though kudos to Sun for the Schematron additions to MSV and for Microsoft for
making sure that MSXML 4 runs Schematron).
Database people will often classify Schematron as not really being a schema
language (because a schema is about static declarations to help storage and
indexing, to them), but they do get the idea fast. Data in a database may
not be a tree; so when we serialize data into a tree for XML and use
a grammar we may lose the ability to express important constraints.
Schematron's path mechanism allows you to express many of the
constraints that grammar-based systems cannot, as well as many kinds of
"application" or "business-logic" constraints.