"It very well may be that, in the future, we’ll look back and realize that microformats were the bootstrap needed to haul the web into semanticity." -- Eric Meyer
Amen to that. The standard wisdom in this space is that you start by digging out pure, pristine semantics and then put them into a formaldehyde solution with the lable:
- Universal form - do not touch.
on the bottle. From these, you generate mere renderings. (X)HTML is only for such throw-away renderings. Second rate, derivative, impure things...XSLT/CSS are prophylactics that insulate the high priests from the possible infections that renderings carry around on their hind legs.
Real XML vocabularies on the other hand, are ones in which domain specific concepts have their own custom tags dammit! Tags I say! Tags are the rightful home of the most hallowed Universals of noun-ness : from cars to coffee pots to chalk to cheese.
He who would slide semantics into (gasp!) renderings under the banner of nasty vomative attribute skullduggery will suffer the death of a thousand first order predicates...
Or so the received wisdom would suggest. Just as received wisdom suggested that the Webs failure to deal with link integrity would render it useless. Just as received wisdom suggested that the only way to find stuff in a large corpus is to have everything classified beforehand. Just as received wisdom suggested that SMS was too limited to act as an effective human-to-human communication tool.
Good arguments all. Simple, elegant and wrong. Every one of them.
Semantics is next up.