httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Paul Sutton <>
Subject Re: Negotiation updates, and transparent neg.
Date Tue, 20 Aug 1996 14:18:55 GMT
On Tue, 20 Aug 1996, Roy T. Fielding wrote:
> Koen's draft sucks, and that's being nice.  It continually astounds me
> how he manages to make things more difficult by breaking the most obvious
> abstractions in the protocol.  Transparent negotiation will include
> content-encodings before it goes anywhere, because two variants of the
> same resource which differ only by content-encoding need to be differentiated
> somehow in order for the rest of his algorithm to work.  In any case,
> its stupid to leave it out.

Good, this certainly seems essential to me.

> No, use 406.  It may not be defined in RFC 1945, but it is a valid HTTP/1.0
> response code and all HTTP/1.0 applications may be capable of understanding
> it without implementing other parts of HTTP/1.1.


On a negotiation related note, can I ask a question about charset
negotiation from HTTP/1.1? It says that all UAs are assumed to handle
ISO-8859-1 -- fair enough, that means we can serve up content we don't
know the charset of, since that is assumed to be ISO-8859-1 as well. Now
say the UA can handle another charset, and lists it on Accept-Charset,
meaning presumably it would _prefer_ the charset it names (otherwise it
would name ISO-8859-1 as well). But content negotiation at present treats
the two charsets as equally desirable. And, more importantly say the UA
can handle two additional charsets, and gives them different q values: say
0.5 and 0.6 (as an example).

Now, my patch considers ISO-8859-1 to have q of 1.0 (since the default q
is 1), so it is always going to be prefered. So the question is: can the
server assume a q value for ISO-8859-1 of below that of explicitly listed
charsets (say 0.001)? Or just give it a low q if the accept-charset line
is not empty and ISO-8859-1 isn't explicitly listed?  Or should UAs
explicitly list ISO-8859-1 with a low q if they prefer other charsets?

UK Web Ltd

View raw message