- From: Koen Holtman <koen@win.tue.nl>
- Date: Fri, 6 Jun 1997 00:04:17 +0200 (MET DST)
- To: Larry Masinter <masinter@parc.xerox.com>
- Cc: koen@win.tue.nl, http-wg@cuckoo.hpl.hp.com
Larry Masinter:
>
>Yes, we were just browsing the mailing list, looking for
>the correction to the problem. If the resolution of this
>issue is that "the BNF is correct, the text is wrong", then
>we need someone to draft new text.
>
>The status should move back to "drafting".
>
>Volunteer?
Just for the record, here is a draft:
Single-character quoting using the backslash ("\") character is not
permitted in HTTP/1.1.
This is the text which is in the 1.0 spec, with HTTP/1.0 changed to
HTTP/1.1.
Koen.
e that the correct 'design' is quoted-pairs are supported.
I would find such a change quite surprising. A stated principle of
HTTP version numbering design is that messages with the same major
version number will parse in the same way. So I would be surprised if
quoted string parsing changed from 1.0 to 1.1.
>This may be a case where the minimal risk of HTTP/1.0 breakage should bow
>to the greater common good.
I don't know if the risk is that minimal: we would need a careful
analysis of current practice to determine that. Skimming the 1.0
spec, I believe that 1.0 only uses quoted-strings in optional mime
type parameters and optional authentication parameters. I don't know
if there are any deployed 1.0 protocol extensions which use
quoted-strings.
One example to illustrate my concerns: if a server is parsing an
Accept header from a user agent which is directly behind a 1.0 proxy,
it cannot tell if the agent is a 1.0 or 1.1 agent. So should it do \"
processing when parsing the Accept header from this agent or not?
>Dave Morris
Koen.
Received on Thursday, 5 June 1997 15:08:46 UTC