On 08/02/06, Stephen Stagg <[EMAIL PROTECTED]> wrote:
> Why do we need an HTML 5? Can't we dispose of HTML and just use
> styled XML in the future? It would be one helluva way to enforce
> standards, and we wouldn't have all this wrangling over exactly which
> element to use.  HTML in itself is not a good example of an XML
> doctype because the paragraph markup does not lend itself to proper
> hierarchic layout.  the heading tags should be able to be subsets of
> a paragraph, for example.

Well, it's a question of attaching semantic meaning to the structure
of the data. XML has zero semantic meaning for elements. In XML,
<hdusdlejncy wiakhjsem="blah"> has exactly the same semantic meaning
as <a href="blah">. That is, no meaning at all. We need some kind of
attachement mechanism for semantics. This is provided in two possible
ways, either externally by the mimetype or internally by namespaces.

XHTML, SVG, RSS, Atom etc. can all be summarised as sets of semantics.
And by all means things closer to the heart of XML such as XLink,
XInclude, XML Schemas, XLS-FO, XLST etc. too.

What we really want to do when we create documents isn't usually just
to provide a structure for data to present in a certain way. We want
to convey some kind of meaning. The meaning can't be conveyed by CSS.
It's possible we could create a semantics attachment model, but
semantics on the whole aren't easily representable for computer
understanding. A much easier solution is to use specific sets of
semantics, which we attach by namespaces or mimetypes. All consumers
can then see if they support the mimetype or namespace, and attach the
semantics of that set of semantics to the underlying structure. In
fact, consumers of XML that don't know the semantic set of a namespace
are still able to say that the meaning is described by that namespace,
even if they don't know in particular what that meaning is.

These sets of semantics are of course the XML applications such as

> The focus would then shift to CSS and the different display-types
> that can be defined for ANY tag.  Microformats and Micro-Namespaces
> could then  be used to allow true semantic delivery.

But really, you need a namespace to attach meaning in XML. XHTML is a
known and widely implemented namespace. Why not use this namespace as
base for extended semantics, instead of introducing new namespaces for
it? And as for microformats, those are actually just extensions of the
semantic set of this very namespace, or extensions of other sets of
semantics. You can't attach semantics to XML without these tools,
really. Microformats are just semantics attached to normally
semantically indifferent constructs in an already existing set of

> I take it this has been suggested before, so what are the arguments /
> counter-arguments ??

Arguments for using plain home made XML is that you might want higher
granularity and specificity of semantics than provided by preexisting
XML applications. But really, to get that you essentially need to
create that set of semantics and assign it to a namespace. Just naming
something <footnote> or <navigation> doesn't mean it gets the semantic
meaning of being a footnote or navigation. Nor does it convey any
particular definition of how to handle that if no presentational or
behavioral hints exists explicitly in the document, because the
defaults on not-strictly-semantical aspects are also part of the
semantic sets (In my view, at least. Which isn't neccesarily canon...)

Counter arguments against it I think I've already mentioned.
David "liorean" Andersson
The discussion list for  http://webstandardsgroup.org/

 See http://webstandardsgroup.org/mail/guidelines.cfm
 for some hints on posting to the list & getting help

Reply via email to