"formatting properties" not ready for consistency

At 01:14 AM 2002-08-16, Roy T. Fielding wrote:
>The TAG finding is not even remotely about transferring representations
>in one format or another.  What it is about is protocol design and the
>difficulty of deploying alternative mechanisms for separating presentation
>from content when each new group responsible for defining those mechanisms
>is allowed to choose arbitrary names for the same concepts.  It is
>reasonable for the TAG to expect, or even demand, that W3C specifications
>consistently use the same terms when they reference the same concepts,
>regardless of the protocol, or at least have a very good reason for
>their departure.

[personal opinion only]

The problem is that it is not just Hakon's request, but the TAG finding too,
which is predicated on the erroneous premise that "formatting properties"
are the same thing from application to application.

"Formatting properties" is an oxmoron, a self-contradictory term.

What we have in CSS and other styling languages are not properties, but
parameters that steer formatting directives.

Formatting is a transformation.  The "formatting properties" as we know them
today are specific to cells that are narrow both in the range space of that
transformation, *and the domain space as well*.  Not just properties proper
to the domain of the screen, but which are proper to the notional taxonomy
of the application domain.

The logical flaw is that while the range spaces are heavily re-used, the
domain concept spaces are all over the place.  There is no de-facto nor
readily-achieved convergence on layers or subdomains, here.  On the
abstract-content side the notions exhibit a "sea of objects" topology in
which there is no re-usable segregation or clustering into notional
neighborhoods in current practice.  HTML elements are too tainted with
presentational presuppositions to serve.

We are on track, if building on the Character Model we can next get a Word
Model and the like up the scale in aggregation to identify notional
ontologies that actually fit what appears in different delivery contexts.
But this is a fond hope for a future clarity of rhetorical classification.
Past practice in the style domain is not close enough for stability now to
be the right idea.

Style rules need to be free to experiment with parameter spaces.

Globally converge the terminology for the properties that are actually
common across applications: color, size etc.  Mostly these are the things
that are observable in the range space alone, and correspond to the types of
the values of "formatting properties," not the names of "formatting
properties."

Recognize that the terms we have today are mostly not universal, but
domain-specific: font-color, button-travel-minimum,
background-beep-how-quiet, etc.  The latter should be addressed in
domain-specific ways.  For the most part it is going to take re-engineering
to separate the domain-portable aspects from the domain-specific aspects
before we have in hand notion sets that are fit for standardization and
stability.

Al

>Hi Håkon,
>
>>Thus, it seems like the principle of separating presentation from
>>content in web documents is well-understood and well-established.
>
>Yes, that should be documented somewhere as a good principle, but
>it is not a constraint on the architecture.
>
>>The recent TAG finding which suggests that XSL FOs is just another XML
>>vocabulary which can/should be stored/transferred on the web breaks
>>with this principle since FOs don't separate content from presentation
>>-- it's all mixed up and one can barely extract the text in
>>machine-readable form.
>
>Just because it is a good principle to separate content from presentation
>doesn't mean the Web should consist only of separated content.  PDF is
>just as applicable for this case, and more frequent in practice, than
>XSL FO's.  There exist legitimate reasons, mainly legal in nature, for
>why some content is inseparable from its presentation.
>
>The TAG finding is not even remotely about transferring representations
>in one format or another.  What it is about is protocol design and the
>difficulty of deploying alternative mechanisms for separating presentation
>from content when each new group responsible for defining those mechanisms
>is allowed to choose arbitrary names for the same concepts.  It is
>reasonable for the TAG to expect, or even demand, that W3C specifications
>consistently use the same terms when they reference the same concepts,
>regardless of the protocol, or at least have a very good reason for
>their departure.
>
>....Roy

Received on Friday, 16 August 2002 12:36:47 UTC