W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > May to August 1996

Re: v11-03 COMMENT: 18.1 Accept

From: <birkettd@vax.cs.hscsyr.edu>
Date: Mon, 13 May 1996 18:21:39 -0500 (EST)
To: Koen Holtman <koen@win.tue.nl>
Cc: Dave Kristol <dmk@allegra.att.com>, http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
Message-Id: <Pine.3.89.9605131823.A541112011-0100000@VAX.CS.HSCSYR.EDU>
	I have no idea what any of this stuff means; I am just a lowly 
medical student in Syracuse, NY.  Why is this happening to me?

On Mon, 13 May 1996, Koen Holtman wrote:

> Dave Kristol:
> >
> >Two questions about the grammar.
> >
> >       Accept         = "Accept" ":" #(
> >                             media-range
> >                             [ ( ":" | ";" 
> >                               range-parameter
> >                               *( ";" range-parameter ) ]
> >                            | extension-token )
> >
> >
> >1) Should that be *( (":" | ";") range-parameter ) ?
> 
> That would also work, but it would even be more horrible.  I have
> received mixed comments about the : stuff at the web conference.  Some
> people like it, some people don't.  The reason for having it is to get
> rid of a potential syntactic ambiguity: some (future) MIME types could
> have numeric parameters called `q'.  Sure, the solution is horrible.
> There is actually another solution, that you may call less or more
> horrible, depending on the parsing technology you use.  See below.
> 
> >2) Is <extension-token> really useful? 
> 
> Yes.  I have use for such an extension-token in my transparent content
> negotiation design.  Of course, I could also put it in another header,
> or even in its own header, but I think it is nicest in Accept.
> 
> > Isn't it syntactically ambiguous?
> 
> No. A media-range always has a / in it, an extension-token never has a
> / in it.
> 
> >(Okay, okay, so it's LR(1).  That's messy, if not unprecedented.)
> 
> >It would be nice to have an example that shows both parameters and
> >range-parameters, such as:
> >        text/html; level=1: q=0.9, text/html; level=2: q=1.0
> 
> Yes.
> 
>  ---
> 
> Other solution to the syntactic ambiguity problem:
> 
>        Accept         = "Accept" ":" #(
>                              media-range
>                              [ range-parameters ] )
> 
>        media-range    = ( "*/*"
>                         | ( type "/" "*" )
>                         | ( type "/" subtype )
>                         ) *( ";" parameter )
> 
>        range-parameters =  ( ";" "q" "=" qvalue )
>                           *(";" extension-range-parameter )
> 
>        extension-range-parameter = ( token "=" token )
> 
>        extension-token = token
> 
> together with the text:
> 
> A media-range which includes a parameter with the attribute name "q"
> MUST be followed by a range-parameters part.  This rule is meant to
> disambiguate the parsing of such constructs.  An example is:
> 
>        text/x-example;q=0;r=3;q=1;r=4
> 
> which is "text/x-example;q=0;r=3" with a quality value of 1.
> 
> In my opinion, It is not yet too late to switch to this solution; I
> could propose appropriate edits to Jim Gettys if many people on this
> list say they like it better than the solution with the colon.
> 
> >Dave Kristol
> 
> Koen.
> 
> 
Received on Monday, 13 May 1996 15:22:30 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:31:59 EDT