- From: Julian Reschke <julian.reschke@gmx.de>
- Date: Sun, 9 Aug 2015 21:31:42 +0200
- To: Martin Thomson <martin.thomson@gmail.com>, Mike West <mkwst@google.com>
- Cc: "public-webappsec@w3.org" <public-webappsec@w3.org>
On 2015-08-09 17:57, Martin Thomson wrote: > On 9 August 2015 at 00:31, Mike West <mkwst@google.com> wrote: >> I'm trying to parse the `Prefer` ABNF (defined in >> http://tools.ietf.org/html/rfc7240#section-2). It says that `token` and >> `word` are defined within Sections 3.2.1 and 3.2.4 of [RFC7230], but they >> don't appear to actually be defined there. Can you point me to the correct >> reference (I assume `token` is from >> http://tools.ietf.org/html/rfc7230#section-3.2.6, but I can't find `word`)? > > That looks like a pretty serious erratum on 7240, which I have opened. > > The "word" rule is not defined in 7230 or even 2616. We might > speculate that word = token / quoted-string, which I have, but all > that means is that you probably shouldn't use it as a model.=. It was in a draft which was current when Prefer was approved. See, for instance, <http://greenbytes.de/tech/webdav/draft-ietf-httpbis-p1-messaging-20.html#rfc.section.3.2.4>. > I think that Julian used a different example of the "canonical" header > field syntax that you could use instead. Another example would have been "Expect", but we removed everything except "100-continue", so it's not a useful example anymore. So yes, "Prefer" is a good example with just word = token / quoted-string added. Best regards, Julian
Received on Sunday, 9 August 2015 19:32:12 UTC