xml-dev - RE: Schema at XML namespace URI to change

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

From: Jonathan Borden [mailto:jborden@mediaone.net]

>The 'problem' with IETF certified MIME types is that each MIME type (or
>group of types) needs to move through the IETF process. 

Control latency means progress is

>A major advantage of namespaces is that they are available to anyone who 
>can create a URI. 

Overcome by an increasing number of tasks. Signal saturation, basically.

>Using the DNS system, a registration and resolution mechanism exists today,
>problems and all, and software (e.g. web server) exists that can resolve a
>URI, parameterized by a MIME type via the Accept: header, into a document.

Into a document or a document handler?  The trick is to recognize a type 
and pass it to a processor for that type.  The problem of XML is that there 
can be many types with a handler per type (potentially) and for namespace 
compounds, the need to open the document to determine which aggregate 
of components are needed.  In VRML, see X3D, the language core handler 
can read a document, determine if additional components are needed, then 
requests these from the system (via the registry) and if not found, 
requests them from the user by asking permission to go get them from 
a known source or alternative.  Negotiation at the level of dialogs...

>Alternatives such as Notations and FPIs have been proposed and specified
>but as of today no pervasive infrastructure exists to resolve an FPI into a
>document. 

Didn't catalogs work?  The problem as I understand it is that FPIs 
and System identifiers work for systems only where catalogs are centrally 
located so resolvable, or ubiquitously distributed.  DNS replicates 
so I am not sure why catalogs can't.  However, one still needs a means 
to automatically add FPI/System pairs to catalogs to be replicated.  
I wonder if spamming isn't the answer.

>So we have a alternative: a system which has problems but
>basically works much of the time, or a theoretically better system which
>hasn't been deployed.

Colonization is hard to beat as a domination strategy.  Seize resources 
and starve competitors until the niche is secured and energy can be expended

on extending control into new niches:  Why HTML Had To Be Simple.

I think that is the "embarassing" part of why it works.  It's hard to 
sustain support for imperialist ethics unless one can convince 
the natives their own best interests are held.  But your point is 
well-taken:  until a better system emerges, we are stuck with the latency 
of the IETF process controls.

len