- From: Koen Holtman <koen@win.tue.nl>
- Date: Mon, 12 May 1997 20:40:57 +0200 (MET DST)
- To: Larry Masinter <masinter@parc.xerox.com>
- Cc: koen@win.tue.nl, http-wg@cuckoo.hpl.hp.com
Larry Masinter:
>
>Yow, now we have to put feature tags in quotes, too?
Feature tags are URIs now like in PEP, so they have to have quotes for
unambiguous parsing.
We could add a condition under which it is OK to omit the quotes, like
in HTML, but optional quotes are not the HTTP way. I have been
thinking about making them optional, but I am a bit concerned about
the additional complexity of lexing/parsing a HTTP header with
optional quotes.
Yet another solution would be for PEP and TCN to specify that URIs
denoting extensions must not contain particular delimiter characters.
I'm pretty neutral on this issue myself. Any strong opinions?
>Look, it was hard enough to push back on
>
>User-Agent: MSIE 4.0, 400x600
>
>User-Agent: MSIE 4.0
>
>to
>
>Accept-Features: "x:screenwidth#600", "x:screenheight#400"
Remember that TCN does not follow the
send-everything-in-request-headers paradigm. We want to push the
above back to
{"home.pda" 1.0 {features "x:screenwidth#0#199"}}
and it is the quotes in this part that we should worry about.
>
>I don't think so.
>
>Maybe we could go back to do/dont/will/wont:
I looked at do/dont/will/wont a long time ago, and I don't think that
it can offer anything to make HTTP negotiation (be it PEP or TCN)
easier.
[...]
>On first principles, I think merging PEP and feature negotiation
>is a reasonable thing to consider.
Well, merging TCN feature tags with PEP extension identifiers is what
introduced the quotes in TCN.
But that is negotiation metadata. I considered merging the
negotiation *protocols* a long time ago, and I concluded that any
merged system would end up being much more complicated than the sum of
the parts. If you think otherwise, please come up with a design.
Koen.
Received on Monday, 12 May 1997 11:45:20 UTC