jettisoning baggage

After all this discussion (1415 messages), I'm coming to the conclusion
that the use of URIs as namespace identifiers was a risky and ultimately
mistaken step.  Today's messages have sounded (even more than usual) like
this list is mad, turning on itself yet again for another spin of the
wheel, to no one's credit or gain.

Rather than enjoying 'modular reuse', we are reopening the details of RFC
2396, trying to figure out how to integrate its prescriptions with the
world of XML documents, and battling over minutiae that seem
ever-more-remote from XML processing in the real world.

This is difficult stuff.  Using it in situations where vocabulary
identification is at stake - not just an occasional broken image or missing
bit of dynamic HTML - seems to have halted a number of critical
specifications for XML developers, while providing very little benefit.

While URIs should provide us with unique identifiers, the one critical
contribution that URIs make to XML itself, even that seemingly simple task
has been complicated by serious questions about reliably determining URI
uniqueness.  The convenience of using the domain name system has been
overshadowed by the extra semantics involved in URIs, in particular the
questions raised by the suggestion many protocols make of retrievable
resources.

I've heard all kinds of wonderful things about the world we will have if
only we use URIs as URIs as namespaces.  I haven't seen any of it made
real, and I strongly suspect that the costs in the short term are going to
infringe on or even prevent the gains in the long term.

All we really need to make namespaces work - and to make them useful as
identifiers of different kinds of information - is a promise of uniqueness,
of ways to compare these identifiers for uniqueness, and best practices
more or less equivalent to "thou shalt not covet thy neighbor's identifier".  

We don't need to identify resources, and we don't need the identified
resources to be retrievable by some process involving dereferencing the
URI.  We only need the 'U' and the 'I', and all the parts involving 'R'
appear to bring confusion and controversy.

So long as we have unique identifiers, we can safely build infrastructure
to support them.  There is nothing sacred about namespaces beyond 'unique
identifiers' and a set of syntactical tools for making those unique
identifiers work in XML.

I suppose that makes me a strong advocate of 'literal'.

Simon St.Laurent
XML Elements of Style / XML: A Primer, 2nd Ed.
http://www.simonstl.com - XML essays and books

Received on Wednesday, 14 June 2000 20:00:48 UTC