On Tue, 14 May 1996, Roy T. Fielding wrote:
> Fine with me. I included appropriate weasel words in the HTTP/1.1
> draft 00 to allow such kludges, but I'm not sure if it survived.
> I'll try to make sure that it is always possible for a server to
> compensate for broken browser implementations.
The current draft spec, <draft-ietf-http-v11-spec-03.html>, says, in
section 18.1 (page 83):
Media ranges can be overridden by more specific media ranges or
specific media types. If more than one media range applies to a
given type, the most specific reference has precedence. For example,
Accept: text/*, text/html, text/html;level=1, */*
have the following precedence:
1) text/html;level=1
2) text/html
3) text/*
4) */*
Which seems correct. Except that if I read this correctly, simply,
say, assigning */* a quality value of, say, 0.2 by default and text/*
a value of, say, 0.5, would not be the proper thing to do. Because if
I read this correctly, consider the following accept header:
Accept: */*, text/html, text/plain: q=0.9
While obviously text/html would (and should) take precedence over all
else, if the choice was between text/plain and image/gif, the image
would take precedence, since it has a 0% reduction in quality, while
the text/plain has a 10% reduction.
At least, that's how I read it. And probably, it's how it should be,
since there may in fact be a rationale for something like:
Accept: text/html, text/*, image/gif: q=0.9
Which would mean give me HTML, then any other text, then a GIF if
that's the only thing left.
--
________________________________________________________________________
Alexei Kosut <akosut@nueva.pvt.k12.ca.us> The Apache HTTP Server
URL: http://www.nueva.pvt.k12.ca.us/~akosut/ http://www.apache.org/
"War does not determine who is right, only who is left."
|