[uf-discuss] Fwd: [whatwg] Predefined classes are gone

[Keith Alexander]

Chris Messina wrote:
>
> This was one of most ill-conceived ideas I can imagine. Leaving class
> values unfettered by the HTML spec is hugely important.
>
> I imagine that the original concept is something of a hat tip to
> microformats -- but avoids the process that we use as well as
> circumvents the peer review of this community. It also seems dictated
> by fiat as opposed to open discussion (but that's a minor point).
>
>

Predefined class names have been dropped, but without @profile (or 
something like it), class names are still effectively predefined by 
whoever is shouting the loudest. We don't need class names decided by 
democracy or autocracy. We just need a standard mechanism that lets 
people solve their own (meta)data-in-html problems and tell user-agents 
how they are doing it without getting in anyone else's way. @profile is 
the closest we've got to that, and maybe it could be improved upon, but 
it would be hugely limiting to the progress of semantic html to drop it 
altogether.

The reasoning for dropping @profile, as far as I can gather, is that,  
in practice microformats, seem to work fine without it. That  may be 
true (for the moment at least),  but  what about  semantic html formats 
that  aren't Microformats?  Microformats 'work fine' without @profile 
because they:
1) cover a relatively small range of attribute values
2) use deliberately obscure classnames
3) have raised a lot of awareness that these obscure classnames have a 
special meaning

Clearly there are lots of people and organisations who would like to 
solve problems by using semantic html, but for whom the Microformats 
approach  to ensuring unambiguity is not a practical option.

Just dropping the functionality of @profile from HTML5 altogether means:

 * You can't use GRDDL with HTML5 
 * No standard way of letting consumers discover semantics behind html 
conventions by "following their nose" from HTML source to profile document
 * It's harder to experiment with new formats and evolve existing ones - 
say you wanted to base your own format on an existing one, there is no 
way to tell a parser that this isn't  eRDF, but a mutant breed of eRDF 
and hCard. And consequently, less innovation will take place in the 
semantic html space.

Conversely, if HTML5 keeps this functionality, and perhaps even tries to 
improve it, and if its use is  well explained, I think  it could  go 
from being a relatively obscure attribute, to being  a fairly important 
feature  of semantic html.  Microformats, after all, has raised a lot of 
interest in using HTML for encoding data semantically, but its not an 
approach that solves every problem. @profile can let people solve 
problems for themselves .

Cheers,

Keith
 

-- 
Keith Alexander
http://semwebdev.keithalexander.co.uk/blog/