Re: Significant W3C Confusion over Namespace Meaning and Policy

On Feb 18, 2005, at 12:36, ext Elliotte Harold wrote:

> Patrick Stickler wrote:
>> And to the recieving agent, it is crystal clear which version
>> of which model to employ to interpret the data instance,
>> and namespaces can be left at the syntax layer where they
>> belong -- and where they do a good job of avoiding naming
>> collisions.
> OK. I think I'm beginning to get your point. I disagree with it, but 
> maybe I'm beginning to understand why you don't like RDDL.
> RDDL is simply not attempting to solve the problem you want to solve. 
> A RDDL document describes a namespace, not a model. I think that's OK. 
> The problems RDDL is trying to solve are perfectly valid problems that 
> arise in practice, and RDDL solves them very neatly.
> Beyond that, I completely disagree at a very fundamental level with 
> the whole goal of exchanging interoperable models as opposed to merely 
> exchanging syntax. Your model used for interpretation of the data may 
> or may not be what I want or need to use to interpret the same data. 
> Models are properly a local decision, not something to be exchanged 
> and legislated.


I presume to have the right to say how data that I create should
be interpreted, and I presume that most consumers of my data
will appreciate being told how I expect it to be interpreted.

Nobody is forced to do anything. You are free to do whatever you
like with my data. But better for it to be clear what I intend,
and that you are doing otherwise, than for it to be left

> Models are fundamentally non-interoperable because different people 
> have different needs and therefore need different models.


> Exchanging the model with the document is an attempt to impose the 
> sender's view of that document onto the receiver;

Not impose, just communicate.

> but the receiver may have very good reasons to want to process that 
> model differently than the receiver wants it to be processed. For 
> instance, think of stock analysts receiving a company's financial 
> data. Why should the analysts accept the company's model of profit, 
> depreciation, stock option expensing and so forth rather than using 
> their own that they find to be more predictive of the market?

I never presumed any mandate. You're reading that into
what I have been saying.

> RDDL is useful because it doesn't attempt to impose processing. All a 
> RDDL document says is, "You seem to be looking for some information 
> about elements in this namespace. Here's some further information that 
> might be useful." Because RDDL places humans front and center and 
> allows them to make the ultimate decision of what is and is not 
> relevant to their needs, RDDL works, for the very small problem it's 
> trying to solve.

Again, I think you are misunderstanding what I am proposing.

> The problem you're trying to solve is much harder, will not be solved 
> by a single namespace document as you point out, and probably should 
> not be solved. That doesn't mean we shouldn't solve the problem of 
> hitting 404s when loading a namespace URI into a browser, though.

Again, I have no problem with folks using RDDL to publish
representations of namespace documents via a namespace name,
so long as the namespace name URI actually identifies the
namespace document.

What I am concerned with, is agents making presumptions about
the interpretation of namespace name URIs which are not
licensed by any specs and developers being confused about
what behavior they can rely on versus what constitutes
optional, localized practice.

Go ahead and use RDDL. I don't mind. If folks find it useful, great!

But let's stay clear about the fundamental architectural
principles and not overstate the scope of such a solution.


> -- 
> Elliotte Rusty Harold
> XML in a Nutshell 3rd Edition Just Published!

Received on Friday, 18 February 2005 15:16:19 UTC