Lists Home |
Date Index |
- From: len bullard <firstname.lastname@example.org>
- To: Tony Stewart <email@example.com>
- Date: Sat, 31 Jan 1998 18:31:51 -0600
Tony Stewart wrote:
> Len Bullard wrote:
> >>It can do what DTDs do well: provide a precise description of the
> presentation style of the interface as a set of routed behaviors.
> I would have thought that a good DTD doesn't do this at all. The DTD
> should define the information content, leaving both style and (IMO)
> behavior to be specified in a stylesheet that is tailored to this
> specific usage of the information.
> Thus, it is the style sheet describes
> the presentation style, not the DTD. Otherwise, how are you going to
> reuse the information in other formats? You're not going to want to
> change the DTD. And you may not have permission to do so in any case.
> Since this is all pretty basic religious thinking, perhaps I
> misunderstood you.
One could say that it is a religious conviction in some cases and
be quite right, and in others, it is an engineering constraint and
be right. It is the *SGML Way*. In that sense, yes, it is a religion,
and for some years, I practiced it. "But what is the good, Phaedrus?"
Look at what you are saying:
1. Stylesheet properties are not "information"
2. Stylesheets express behaviors. So in fact, a stylesheet
language is a programming language, Turing complete if you will.
3. For some kinds and instances of information, there are lifecycle
requirements for reuse.
4. For some kinds and instances of information (DTDs in your example),
there are policies for the behaviors that can be applied to the
kinds and instances of information.
1. I don't think you intend one. But it is often a hidden premise in
the debates about separating style from content (which is what you are
using information). That distinction proves to be thin. Perhaps
by stylesheet information, you mean, typographic properties.
2. Stylesheets that express behaviors are simply programming languages
with structures (data types) for typographic properties. In this
view, Java/AWT et al is a stylesheet language. After that, choosing one
comes down to practical engineering requirements of platforms,
interoperation with other engines, etc. Anyway, in this view, VRML
is a stylesheet language. Perhaps the best way for it to include
text support is to include it natively. This idea has come up and
there is a text node in VRML which browsers like WorldView can display
very well. (NOTE: The issue of reformulating VRML as XML is one
of the framework efficiency, not descriptive power or lifecycle.)
3. This is true of course. But unless requirements are very carefully
examined, no size fits all.
4. True and it varies widely. One of the features of DTDs that make
very attractive for policy is the ease with which they can
be adjusted liberally on site of use. This one slips by most of the
SGML theorists who do not work in production sites where multiple
DTDs are used at different points of a process or procedure. In other
they are an instrument of policy, not a policy. Information is not
where a high rate of change prevails. A DTD is more like a control knot
NURB than a point in a B-spline.
My point is that for many information engineering problems, the approach
Pierre took with Prototype has been taken by others and successfully.
The arbiter of success is not the religion of the SGML Way, but the
ability to meet the requirements of the task. Bytes aren't holy.
As XSL/XML/XLL reach ever greater levels of design complexity in the
base standards, a question emerging in other design groups (one heard
before during the HyTime/DSSSL era) is: Are these really complicated
solutions looking for problems, not new and vital technologies? Is
there sudden rush of popularity based on the soundness of applicability,
or the product of software company juggling of public perceptions?
If simpler and more readily available and more easily understood
technologies exist to solve a problem with an acceptable timeframe
exist, the experienced engineer and the practical customer adopt
them. If not, they try the next best thing. Is XML a *religion*
of just the next best thing?
xml-dev: A list for W3C XML Developers. To post, mailto:firstname.lastname@example.org
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/
To (un)subscribe, mailto:email@example.com the following message;
To subscribe to the digests, mailto:firstname.lastname@example.org the following message;
List coordinator, Henry Rzepa (mailto:email@example.com)