[whatwg] Generic Metadata Mechanisms (RDFa feedback summary wikipage)

[Ian Hickson]

On Wed, 10 Sep 2008, Manu Sporny wrote:
> 
> There will, of course, be many more examples of the problem following 
> the same format as shown above. Is this what you had in mind for the 
> problem description? If so, give us some time and we'll be able to 
> refine that page in the coming months.

That's indeed the kind of thing that would be helpful. It's also important 
to indicate why we think that authors will want to actually indicate the 
information here. For example, in the music case, do authors want to 
expose that information, or are we merely hoping they will? Ideally, 
evidence of browser vendors, browser extension authors, or authors working 
around the lack of a feature is the kind of evidence we need to 
demonstrate a need. For example, are there browsers that are implementing 
<audio> in a way that hoks into iTunes? Or are there extension authors 
that support a custom class attribute that interacts with WinAmp? Or has 
Amazon looked for a mechanism to transfer purchases into media libraries 
other than using their download tool? Would they actually use a mechanism 
if we provided it? How would such a mechanism work, and would it be 
well-defined enough to handle things like resumable downloads, financial 
transactions, and individualised MP3s interoperably?

Basically we need to show that we are addressing real needs and not just 
abstracting out what looks to us like common elements to multiple problems 
and providing a single generic mechanism that doesn't actually usefully 
solve the real problems.

The same reasoning is why HTML has a bunch of explicit data types for its 
form <input> controls instead of having a generic type system like XML 
Schema and letting <input> controls be invented for any type. While the 
latter, with user agents dynamically creating form controls on the fly, 
would be a much more generic solution and would satsify the computer 
science abstraction aesthetic, it turns out to really not be what authors 
want, and at the end of the day an explicit but closed set of types ends 
up satisfying the bulk of authors much more. Although it doesn't solve all 
the problems that a generic system does, it solves the 80% case better, 
and that is what matters.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'