Another look at namespaces

[Tim Bray]

At 09:18 AM 9/17/99 -0400, Tim Berners-Lee wrote:
>From: Tim Bray <tbray at textuality.com>
>>Among other things, I don't believe that most
>>interesting namespaces *have* definitive information, but have semantics
>>that are communicated via some messy combination of schemas, stylesheets,
>>prose documentation, and running code.
>
>We either have a different use of words or a very serious
>problem.  Whereas with natural langauge, meanings change and
>grow and everyone has slightly different associations with a word,
>in computer languages we need to build on top of XML we need
>to have the ability to define meaning precicely in terms of
>other existing languages.

Actually, Tim B-L and I are both advocates of moving as much semantics as 
possible from idiosyncratic impenetrable procedural code and woolly forests 
of prose into declarative schemas so that you can start to use some 
knowledge rep tricks and do a bit of meaning-discovery, not just the 
resource-discovery enabled by the current Web.  This is what Tim calls 
"the semantic web" in his platform speeches, and its foundation is
RDF and RDF style schemas.  I'm onside with this.

Perhaps the disagreement (if any) is on the degree to which this
is possible.  There are going to be lots of languages invented, and
while I believe that we will get increasingly better definitional
machinery as time goes on, I see no reason to believe that any of
the following are going away:

- human-readable documentation
- messy procedural code that actually does something useful with the data
- stylesheets (plural)
- related resources containing hyperlink networks
- related resources containing RDF assertions in some vocabulary
- transformation specs along the lines of XSLT

and so on and so on.

What bothers me is the notion that among all these things, the schema
is somehow privileged and special.  I think that which of the above
laundry-list you care most about is very highly application-dependent,
data-dependent, and rather wholly unpredictable in the general case.
I suspect that the proportion of real-world applications that actually
use the schema at run-time will be moderate at best - schemas come more 
usefully into play in application design, authoring support, and so on.  
Perhaps Tim B-L thinks that this proportion will in fact be quite high?

This is a reasonable disagreement, and we've all been fooled by
technology enough times to have learned humility.

So.......... where do we still have disagreements with concrete
short-term implications?

1. I'm convinced that the namespace mechanism, while it's useful
   for what it is, is hopelessly inadequate as a tool for direct
   mapping between instances and schemas.  We need at least one level
   of indirection, i.e. the packaging work that's hanging fire in
   the W3C XML activity.
2. For this reason, I think that the correspondence between
   namespaces and DTDs in XHTML 1.0 PR doesn't do a good job of meeting
   the needs of either people who need schema discrimination or of
   people who think HTML-is-HTML.
3. I am astonished and disappointed that the W3C still can't bring
   itself to publish a 3-line note saying that for those who see 
   HTML just as HTML, and who want to mix & match those tags with 
   others, they should use the following URI as a namespace name for
   HTML: <insert any old plausible namespace URI here>

 -Tim

xml-dev: A list for W3C XML Developers. To post, mailto:xml-dev at ic.ac.uk
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/ and on CD-ROM/ISBN 981-02-3594-1
To (un)subscribe, mailto:majordomo at ic.ac.uk the following message;
(un)subscribe xml-dev
To subscribe to the digests, mailto:majordomo at ic.ac.uk the following message;
subscribe xml-dev-digest
List coordinator, Henry Rzepa (mailto:rzepa at ic.ac.uk)