RE: ATOM representation of search result facets

Hi Erik,

Thanks for the discussion.  It's really useful to me, hopefully it helps
others thinking about the same thing.

> > That doesn't, of course, make it not useful.  HTML is 
> extended every 
> > which way, and typically no signposts are added to the message to 
> > indicate the nature of the extensions.  It might be useful 
> in feeds to 
> > indicate that they are georss-enabled, and which profile of georss, 
> > for example.
> > But IMHO, such extensions are not in the 'architectural 
> style' of the web.
> 
> look at profiles as a convention of how to signal that 
> clients might expect something. to some extent it just means 
> there's an easier way to find out that a representation 
> follows some additional constraints and extensions, that's all.

I agree it consolidates useful information into one place; my concern
is that in making a standard link relation for how to convey overall message semantics, 
it undermines the role of the media type in web communications.  

> 
> it seems that what you're saying is that in order to be 
> properly RESTful, each and every extension needs to mint a 
> new media type.

A parameter would be preferable, as was done by AtomPub 'type'.

> for well-designed extensible media types 
> (such as atom), i would claim that this actually would be a 
> bad idea. you would have a media type for plain feeds, one 
> for GeoRSS feeds, and one for podcast feeds. and would you 
> also have a media type for GeoRSS-enabled podcast feeds? and 
> all other possible permutations of all extensions? 

I *like* the notion of using a standard media type parameter to
convey this information.  It does not obscure the original 
media type, and can / should be ignored by processors which
don't support the parameter.

> and as long as these things are 
> feeds (i.e., can be used as feeds and behave like feeds), i 
> would argue they should be served as feeds.

I agree; consider: application/atom+xml;type=feed;profile="georss-gml"
it can still be used as an atom feed.

Cheers,
Peter