Re: RDF namespace conventions

At 12:05 PM 2000-05-22 -0400, Simon St.Laurent wrote:
>At 11:51 AM 5/22/00 -0400, Tim Berners-Lee wrote:
>>So, the layer separation you suggest would work only with either
>>(a) no relative URIs -- at least a warning that XML lower layers don't grok
>>them, or
>>(b) change lower layers to absolutize before comparing.
>>
>>Either of these would be consistent.  The second would be cleaner.
>
>Neither of these is necessary.  You're driving the quest for 'consistency'
>too deep into the underlying layers.

The warning option under (a) does not require any lower-layer changes or
awareness, so I cannot follow what you mean as "driving too deep," on this
choice.  Option (b) is clearly invasive and disruptive.

There does not, as you have noted elsewhere, need to be enmity between
inter-layer distinctions and trans-layer connections.  In fact we can stick
to the morality of "sticking to what we said" precisely because the
difference between (a) and (b) has so little actual effect on the
trans-layer connections.  It is the integrity of those connections that the
"semantics first" morality is about, not whether the implementation has a
distinguished syntactic layer or not along the way.

>
>XML processing does not require either of these approaches.
>

Please be careful how you use the phrase "XML processing."  Processing XML
syntax, even with namespacing, does not require any of the above.  But the
applications that justify using multiple namespaces in a document do, in
the main.  Attaching the epithet "XML processing" to one and not the other
of these layers is gratuitously divisive, an irritant we can well do without.

As a counter-balance to the "stick to what we said" morality, we should
also be considering the morality of building polyglot XML usage out of
modules which _unite_ semantic connotation with syntactic designation.  My
toy example is blind access to tabular information, but this is just the
one that is easy to walk through fast.  The
potential benefits from XML as syntax+semantics are endemic.

XML is not formal language.  XML is markup language.  It is a syntactic
toolkit for building markup dialects, and these dialects are _semi_-formal
language, where the semantic alignment between the formal structure of the
markup and natural language sense of the entrained informal language is
_very_ important for all users, not just those who must morph message data
into unexpected presentation views.  

Building for an XML future where XML modules routinely define
syntax-to-semantics bindings generates social value added to a degree it
must be considered when one enters a moral spreadsheet for technology
planning and development.  It is not just a separable parallel track.
Layering semantics over the syntax is an extremely useful framework for
analyzing the technicalities of implementing capability growth through an
accretion of incremental documents.  But moral accountability sits at the
bottom line where the value added is measured in terms of semantics, and
syntax is an instrumetality, morally indefensible by itself. 

In the evolution to modular markup usage, there needs to be a broad
understanding that essentially all modules will have some defined proper
semantics, and that this is generally more than can be recognized from the
syntax alone.  The scoping of modules needs to be driven by the semantics,
and be implemented in a layered fashion which, it appears, does not need to
tear up the presently laid lower layers one whit. 


XML 1.0 is light on semantics.  That is not a virtue, it is a calculated
risk.  

Tim's vision is [morally] right.  Let's see how we can, in all civility
with regard to past syntactic commitments, make speed in that direction.

Al

Received on Monday, 22 May 2000 16:02:42 UTC