RE: [namespaceDocument-8] 14 Theses, take 2

On Mon, 2002-02-18 at 16:29, David Orchard wrote:
> I believe we must solve this definition problem and document this in our
> arch document.  We obviously have 2 camps:
> 1) schemas are syntactic validation, and dtds/xml schema/relax/dcd/WSDL are
> instances.
> 2) schemas are any kind of definition used in interpretation.  This includes
> 1) but adds in html, RDF Schema, others.  A fragment from RDF Schema
> illustrates "Unlike an XML DTD or Schema, which gives specific constraints
> on the structure of an XML document, an RDF Schema provides information
> about the interpretation of the statements given in an RDF data model. "
> Is there any possibility that the proponents of option #2 could see their
> way to adopting schema definition #1?  I would volunteer to help with coming
> up with a new term for #2.

If it were that simple, I'd agree to stop using 'schema' for 2).
But even in the case of XML schemas, I'm not sure folks
see them as just about syntax.

Tim Bray said something like "the world thinks schemas are
for expressing syntactic constraints, so if
we decide otherwise, we'll have to re-educate them."

> I venture that every developer in my company and probably almost
> all of our > 10 000 customers consider #1 to be the correct definition.

That's not what I see.

Visit Robin Cover's "what's new in XML" on any given day, and you'll
find at least a handful of items like this:

* [January 18, 2002]   ARTS and IXRetail Release XML Price and Digital
Receipt Schemas for Retail Industry.     Recent announcements from the
Association for Retail Technology Standards (ARTS) and the International
XML Retail Cooperative (IXRetail) describe the release of new XML
schemas for use by retailers. The IXRetail digital receipt XML schema
"provides retailers the ability to create, deliver and archive digital
receipts using in-store point-of-sale transaction data. ..."


They refer to the product of their labors as 'schemas'.
And this product is *not* just an agreement about syntax,
as far as I can tell.

In particular, if you took their schema and made a copy of
it with a different namespace name, the syntax would be
nearly identical -- documents conforming to it would
have the same size and shape. But they wouldn't be
considered documents in the same language at all.

Languages are more than just syntax. When folks
talk about schemas, they're talking about language
definitions, including both syntax and semantics.

Hmm... maybe that's a useful term or collection of terms:

	language definition -- a (written) agreement about
	some syntactic constructs (aka a set of documents)
	and what they mean.

	syntax specification -- a document that
	includes enough information to tell whether
	some other document is syntactically good or not;
	perhaps including enough information to make
	a parse tree and decorate it while you're deciding.

> [...] it seems
> important to be precise and also obey the least-astonishment rule.

I completely agree. But I think the ARTS folks would be
surprised if you told them their schemas are just about syntax.

Am I completely off the mark?

I don't mind reserving the term 'schema' to mean
something syntactic in TAG use, if that's really
the way it's used in the wild. I just wonder
if that's the way it's used in the wild.

Dan Connolly, W3C

Received on Monday, 18 February 2002 22:31:21 UTC