Re: AW: XHTML: Combine p and section? A few more hints.

On Wed, 29 Dec 2004 20:18, you wrote:
> I don't want <ps> to be "a paragraph or a section". I think it should be =
> "a structural group of text", because both <p> and <section> group text =
> in a similiar way. The only difference is that <p> is the lowest-level =
> grouping, and <section> builds all the other levels.
> Of course, if you think the semantics of a paragraph are not only to =
> group a bit of text but also anything else, we should keep <p>.

That seems illogical from where I sit.  I think the semantics of a paragraph 
_are_ to group a bit of text, which is exactly what they do now.  The real 
meaning of a paragraph, paraphrasing heavily from the real definition, is "a 
collection of sentences which represents a single/contiguous thought."  
Whether it actually gets rendered as a block of text... that's up to the 
stylesheet, really, but the majority of people will render it as a block of 
text since that's what people are used to seeing on the screen.

> I didn't want to use XPath inside of CSS but I meant that a browser =
> could internally use XPath to determine the difference between =
> paragraph-like <ps> (rendered as paragraph (e.g. with margins)) and =
> section-like <ps> (normally no special rendering, but gives ability to =
> find out about the structure) without the need of much extra code.

Ah, but the question then is, how do site authors style a paragraph versus a 
section?  The only real benefit of CSS is that it allows complete 
customisation, and if you make this an internal feature of browsers, then you 
lose something which would otherwise be perfectly customisable through CSS.

Incidentally, I have no major objections to an entirely block-based mark-up.  
I just don't see its place in HTML since HTML is trying to achieve different 
things.  For the king of block-based mark-up, see XSL:FO.  List items and 
even lists themselves can all be represented by a single element, given 
enough attributes.  XSL:FO almost seems like it could do well as a semantic 
markup language, if only it happened to have attributes which dealt with 
semantics instead of presentation. :-)

It's good to have simple rules to differentiate sections of the document 
though.  For one, the CSS required to auto-number a set of nested <section/>s 
is much simpler than what you'd need if paragraphs had to be marked up with 
the same tag.

TX

-- 
             Email: Trejkaz Xaoza <trejkaz@trypticon.org>
          Web site: http://xaoza.net/
         Jabber ID: trejkaz@jabber.zim.net.au
   GPG Fingerprint: 9EEB 97D7 8F7B 7977 F39F  A62C B8C7 BC8B 037E EA73