Re: Shelby's Final Position Paper on XBL

Hello Shelby, thanks for the summary.

I do not make any statments about XBL here. I don't know enough of XBL to feel 
competent enough to do so. And since we have quotas here I keep only 1-2 
weeks of list mail. I don't want to change my quotas just because someone 
isn't able to make it short (neither am I, as you can read in this post).

But I know enough of XML/XSLT/CSS/TeX/LaTeX/DocBook/XSL to break up, tear and 
shred your complete chain of argumentation because imho you assume wrong 
definitions of the important terms.
That does not mean that your conclusion is wrong. But it sounds to me a bit 
like "If she weighs the same as a duck, she's made of wood and therefor is a 

Am Montag, 6. Januar 2003 14:53 schrieb Shelby Moore:
> <h1>Shelby's Final Position Paper on XBL</h1>
> <h2>What Is Style?</h2>
> <p>Seems to me that "style" in W3C context, is separation of "presentation"
> from "markup".</p>
Wrong. I don't say my definitions below are better, but yours is wrong. Sorry.
I consider most of your definitions wrong.

Markup is just the use of tags to markup content. Markup is neutral, markup 
can be presentational, logical or whatever.

Style is defining how presentation works.
Look at XSL. Can you see any seperation of presentation and markup there? 
Still XSL is Style.

Presentation is presenting content to an intelligent user, which usually is a 
human end user only, but which also might be cats, dogs, monkeys, aliens 

Style can be associated to content for creating a presentation and needs to be 
associated to content if its markup does not imply any or too little style at 

So presentation is the result of combining content, its markup and the style 
associated with the markup.

Semantics is the meaning of something in a certain context. Semantics can be 
specified by a specification describing the semantics of markup for a certain 
markup language. These are logical level semantics. Semantics can also be 
implied by associating a certain style with markup. These are presentation 
level semantics.

Sometimes the semantics of a language are style itself (see XSL and CSS, where 
CSS of course is not a markup language).

Presentational markup is specifying style, specifying how something look, 
sound, feel like.
Semantical (Logical) markup is specifying semantics, expressing what you mean.

Semantics and Presentation are combined in the way that Presentation is often 
used to imply certain semantics.
(The well known crux is that when only the presentation is used to imply 
semantics, so only presentational markup is used, it is very hard for 
non-intelligent user agents to detect semantics)

Both, logical markup and presentational markup, can have predifined or 
preassociated style. In logical markup, the preassociated style is only used 
as a helper (consider (X)HTML/CSS em { font-style:italics; }), while in 
presentational markup the markup and the style are very closely related to 
each other (consider (X)HTML/CSS  { font-weight:bold;}) or even fall together 
to be one (consider XSL). Logical or structural markup may also have no or 
nearly no associated style (consider span and div in (X)HTML or consider 
XSLT, which has no associated style at all, or consider RDF).

Behaviour is ((a dynamic extension of (style or semantics)) or (some 
specification)???) which implies a certain degree of interactivity above 
style and/or semantics.

I think we have far too little experience with behaviour to decide wether:
- behaviour is an extension of style
- behaviour is an extension of semantics
- behaviour is something new next to style and semantics
- behaviour can or must be seperated in presentational behaviour and semantic 

If you consider a layer model:
* Presentation
* Logic
* Content
Where is Behaviour? It is crosscutting. It is an aspect, extending and 
crosscutting the layers.

I consider the a element of (X)HTML implying some behaviour.
I also consider most or all of the elements of XSLT implying some behaviour.
I consider most of CSS being somewhat static.
I am looking forward for those parts of CSS3 which cover at least some 
behaviour because currently the behavioural parts of (X)HTML can not be 
reimplemented using XML and CSS, but I'm keen on the elimination of the 
logical description of XHTML by reducing it on XML with a DTD or Schema and a 
default CSS style sheet, though I don't know wether this will come.

I don't think CSS is the wrong layer for adding behaviour to a document at 
all. It can't be the correct layer for adding behaviour markup to a document, 
because CSS neither is markup itself nor does it add markup.
CSS is the wrong layer for semantics if the intention of CSS must only be 
style. But is behaviour always semantical?
I definitely think CSS is the correct layer to add behavioural parts since 
there is no other layer until we create a new one and experience with 
behaviour is too little to well define and create such a new layer (and, as 
we know, behaviour is a cross cutting concern).

Yes, XBL is at least in parts W3C redundant. Parts of XBL could be done with 
CSS, parts of it could be done with HTC, parts of it could be done with XSLT.
But on the other hand, they also complement each other quite well.

I do not already have made up my mind wether XBL is good or bad, and I'm sure 
I won't do that soon. Still, XBL is just a Note, not a Recommendation, it's 
an idea formed as kind of little spec submitted to and published by the W3C, 
just as, a Note.

> <p>One could argue that presentation is thus any thing which is not
> markup.&nbsp; That would seem to be overly broad, because then there would
> only be two working groups at W3C, style and markup.&nbsp; This would mean
> for example that XEvents group should be merged under the style working
> group, because events are not markup.&nbsp; Or that if XEvents is markup,
> then it should be merged under markup group.</p>
Too vague.

> <p>So what is presentation?&nbsp; I have argued that presentation is any
> implementation of markup which creates semantics that are compliant with
> markup specifications.&nbsp; Most (if not all) of CSS fulfills this
Wrong. Presentation does not create semantics, it implies semantics.
And CSS is not a markup language at all. You don't use CSS to markup. CSS can 
be used by markup of other languages.
I imply this by reading the CSS 2 spec. The term markup is avoided. The term 
markup is only used for HTML or XML, not for CSS.

> presentation test.&nbsp; And I have argued that any implementation of
> markup, which creates semantics that are <b>not</b> compliant with markup
> specifications, is thus not presentation.</p>
This is completely confused and wrong.

With such a confused start and mix of terms with wrong semantics I can't read 
on because it screws my mind in a negative way. I'm not against thinking in 
strange orbits, but I'm against senseless torturing of well-known 

I would really appreciate you rewrite your final position after rethinking my 
trys of defining the important terms like style, semantics, markup, 
presentation, behaviour.
You may publish it on your web site and just send the absURL.

Christian Wolfgang Hujer
Geschäftsführender Gesellschafter
Telefon: +49  (0)89  27 37 04 37
Telefax: +49  (0)89  27 37 04 39

Received on Monday, 6 January 2003 11:23:06 UTC