Re: PaceSimpleContentType

[Robert Sayre <mint@xxxxxxxxxxxxxxx>]

Ken MacLeod wrote:

> Robert, you misunderstood Graham's point.
>
>  
No, neither of you understand the (very common) use case.

> Graham's point is [appears to be] that the above sample has an Atom
> extension called <FlashParams> that apparently a client should "pass
> to" the handler of the content/@src.  How should Atom spec that?  
Atom shouldn't spec it. If the producer thinks <content src="..."> is 
not useful to an aggregator unaware of the extensions, they can omit 
<content> altogether. This is exactly what they will do if <content> is 
required to be HTML. Feed producers would have to include "100% 
redundant" content in order for general purpose aggregators to 
understand it.

> for
> a variety of potential content types and extensions?  Graham's point
> is that X/HTML already does spec that, so why not have the content be
> X/HTML and let *it* link to the flash?
>  
Because there are two handlers here:

Handler 1: Atom Consumer
Handler 2: HTML Component (WebKit, MSHTML, etc)

Putting information in the object tag passes that information to the 
referenced SWF via the Flash Player via Handler 2. In this use case,  
there is a Flash Movie playing the role of Handler 1, and the metadata 
is intended for it, not the referenced SWF. The SWF in the role of 
Handler 1 shouldn't have to parse html:object and its arbitrary X/HTML 
child content (including more html:objects) to get the metadata it 
needs, especially if both sides of the wire aren't controlled by the 
same people.

Robert Sayre