[whatwg] Semantics in HTML (was: Re: Joe Clark's Criticisms of the WHATWG and HTML 5)

[Leons Petrazickis]

On 11/1/06, Anne van Kesteren <fora at annevankesteren.nl> wrote:
> On Wed, 01 Nov 2006 20:55:58 +0100, James Graham <jg307 at cam.ac.uk> wrote:
> >>> And HTML5 isn't that semantically pure anyway.
> >>
> >>  Where can it be improved?
> >
> > To take a slight detour into the (hopefully not too) abstract, what do
> > people think the fundamental point of semantics in HTML is?
>
> I've no idea.
>
> [snap]

The fundamental point of semantics is to:
- make document and application creation easier
- make document and application maintenance easier
- make documents and applications more parseable, readable, and
reverse-engineerable

When our notions of best semantics conflict with one of those three
tenets, we need to step aside and reconsider our axioms, our
architecture, our design, and our approach.

HTML is a lot of things, and one of the main ones is HyperText Markup
Language. We are marking up HyperText -- we are marking up the web. We
need to mark up things that are on the web, that have appeared , that
are coalescing, and that are just budding. We need to understand what
the concepts are and where the natural boundaries lie, and create
markup that matches natural boundaries. Natural boundaries are not
always consistent -- we need to strive to match natural boundaries
rather than to be consistent.

Do people think in terms of articles? Then we need an article tag. Do
people think in terms of comments? Then we need a comment tag? Do
heading tags conceptually match post titles or entry tiles or comment
titles? Maybe they do. Do we have a way of marking up authors, dates,
times? Do we have a way of marking up "tags" and categories and
"folksonomies"? If we don't, then maybe we should think of things in
that regard, because free-form keyword categorization is a natural
concept on the web.

And natural concepts, boundaries are what matters. We shouldn't
subdivide any more than people intuitively subdivide.

Regards,
--
Leons Petrazickis
http://lpetr.org/