Re: charset Reality Check
Dave Kristol (dmk@allegra.att.com)
Wed, 11 Jan 95 10:49:23 EST
I've been patiently watching the discussion of Accept-Parameter, et
al., and waiting for someone to mention my extensions proposal
(http://www.research.att.com/~dmk/extend.html) as a possible approach.
Since no one did, I will.
I hate to see HTTP cluttered with a bunch of rarely-used headers. At
the risk of using more characters (Roy Fielding's caution noted),
Accept-Parameter could be rendered (from the client):
Extension: HTTP/charset required=oneof; set=UNICODE-1-1-UTF-8
Extension: HTTP/charset required=oneof; set=iso-8859-*
The server would either respond with a similar header, like
Extension: HTTP/charset set=UNICODE-1-1-UTF-8
to identify what was used, or it would return a failure, because neither
of the required sets was supported.
An Extensions:-unaware server would ignore the header and return none,
and the client would have to fend for itself, as it does now.
Roy Fielding said:
> >> Also, a quality attribute is
> >> only useful if we add it to the content-negotiation algorithm --
> >> something I would like to avoid (like the plague that it has become).
> Pre-emptive content negotiation just doesn't work. Perhaps we should try
> to define a matrix of common client characteristics/behavior, and just
> provide a header for that....
The matrix approach is interesting (rather than negotiate each little
item). My Extensions: proposal includes a negotiation algorithm. I
think it would be better to settle on a single algorithm than have a
different one for each thing to be negotiated. Perhaps the one I've
proposed is unsatisfactory. I'm open to improvements. Nevertheless, I
think there will be things in HTTP that will demand negotiation.
Dave Kristol