Re: XHTML2: Proposal for total separation of semantics from structure

On 9/27/05, Junk Account <avoid.spam.account@gmail.com> wrote:
> hello, david
>
> Please see
> http://www.w3.org/TR/2005/WD-xhtml2-20050527/mod-core.html
>
> Either I miss somehting, or it says that class *could* be used ("for
> instance") for css. But it *does* say that is generic. Not for styling, not
> for meaning, but simply to specify some elements (and not others) as
> belonging to a same group, or class: i.e, is general purpose.
>
> After re-reading it, I wrote this:
>
> If you ask me, it means you can not blame web designers much for mixing up
> meaning and style.
>
>
> Please consider the following ponts:
> 1) Web designers did not have, at least up to now, other way to do their
> styling. Particularly, they did not have a styling-specific specification
> mechanism. The class atribute, the only one provided, was a generic one. In
> short, they did not have a style="" atribute.
> 2) There was, up to now, not a quite compelling reason why this should not
> work, or cause any problems. Why not? For instance, beacause the semantic
> web vision was not quite there yet. However...
> 2.1) The semantic web vision is getting ready for prime time, and is a
> massively important development.
> 2.2) XHTML 2.0 is also a big change, and a very rare strategic oportunity,
> since it breaks backwards compatbility. That does not happen every day,  nd
> we should take maximum advantage.
> 2.3) Both things are bound to happen at just about the same time.
> Which leads us to...
> 3) Proposal for total separation of structure from semantics.
> Which also lead us to...
> 4) Proposal for creation of styiling-specific and semantic-specific
> specification mechanisms (i.e, style="" and meaning="", or similar). In this
> context, I sustain that class="" should be mantained for the same reasons as
> before: extensibility, given that we can not foresee all other possible uses
> today.
> If you noticed I just asked in this same list, if XHTML copy/pasting *with*
> formatting (and potentially meaning) could be achieved. The very encouraging
> answer was "yes", it can be done with XHTML in its present form.
> However...most pages today are styled using class="", which is a generic
> mechanism. If we wanted to implement algorithms for collecting all style
> information and all meaning information when copying and pasting
> XHTML...wouldn't we run into trouble? Wouldn't we need styling-specific and
> meaningspecification mechanisms specific for styling, and for semantics?

The only issue I have is that class was the perfect name for this
attribute. Subclass would also work. There are many different
relationships data can have and in this particular case, the
relationship we're trying to use is "is-a"; often called class or
classification. A body of text can be classified as a paragrah; it is
a paragraph; it cannot be a body of text has the meaning of a
paragraph. So while the sentiment is good, in my opinion the name of
the attribute has to be carefully choosen to reflex this relationship.

--

Orion Adrian

Received on Tuesday, 27 September 2005 12:40:24 UTC