Re: [RFC] HTTP Information Request

In the linked message, you say:

> I think we should deprecate HTTP content negotiation, if only to
> make it clear to people reading the spec that it doesn't really
> work that way in practice.

Seems like some explanatory text, at the least, might help people  
understand this feature a bit better.


On 22/06/2007, at 3:27 AM, Larry Masinter wrote:

>
> This proposal seems to fall into the same trap that most proposed
> HTTP extensions fall into: there's no motivation to deploy this
> in clients because most servers don't support it, and no motivation
> to deploy this in servers, because most clients don't support it.
>
> Unless you have a better story for how this will get deployed,
> its mainly an academic exercise.
>
> Things might have been different when HTTP 1.0 or 1.1 were
> being developed, but that's not the case now.
>
> That's the general problem. The specific problem with this
> is that it's a kind of reverse content negotiation, and many
> of the features you're thinking of (e.g., screen/window size,
> accessibility requirements) fit into the framework of media
> negotiation, and the others might, with a bit of stretching
> (e.g., "timezone" as a media feature meaning "content
> appropriate for someone in the named timezone", or, more
> likely, locale.)  In most cases, we talked about the combination
> of client characteristics, capabilities and preferences,
> which seems to cover almost all of your tokens.
>
> There's been a great deal of work in this area, most of
> it not deployed (for reasons above), e.g.,
>
> http://www.imc.org/ietf-medfree/ in IETF and
> http://www.w3.org/TR/CCPP-ra/
> http://www.zurich.ibm.com/ucp/
>
> In general, media negotiation in HTTP hasn't been successful,
> see note & following discussion:
>
> http://www.alvestrand.no/pipermail/ietf-types/2006-April/001707.html
>
> Larry
> -- 
> http://larry.masinter.net
>
>
>


--
Mark Nottingham     http://www.mnot.net/

Received on Friday, 22 June 2007 03:32:13 UTC