- From: Nico Williams <nico@cryptonector.com>
- Date: Sun, 10 Feb 2013 18:16:02 -0600
- To: Poul-Henning Kamp <phk@phk.freebsd.dk>
- Cc: Zhong Yu <zhong.j.yu@gmail.com>, Julian Reschke <julian.reschke@gmx.de>, "Martin J. Dürst" <duerst@it.aoyama.ac.jp>, James M Snell <jasnell@gmail.com>, "ietf-http-wg@w3.org" <ietf-http-wg@w3.org>
On Sun, Feb 10, 2013 at 6:12 PM, Nico Williams <nico@cryptonector.com> wrote: > On Sun, Feb 10, 2013 at 4:52 PM, Poul-Henning Kamp <phk@phk.freebsd.dk> wrote: >> Did I misunderstand that ? > > I don't know. I wasn't there. But where HTTP/2.0 carries text in the > headers, that text must either be US-ASCII only, or UTF-8 only -- no > other codesets, encodings, nothing. Anything else will result in > just-send-8/just-use-8 and associated interop failures OR de facto > agreement to just-send-UTF-8. The latter would be fine, but better > make it de jure. If this was decided incorrectly in the past then the > decision needs to be open to review: I guarantee this will come up in > IETF and IESG review. Better get it right now. I suppose I should qualify this by saying that header text slots where other standards apply, such as hostnames (IDNA) and IRIs, the protocol can leave the matter to those standards. But that need not be trivial either: for domainname slots we might need to say only A-labels apply. Nico --
Received on Monday, 11 February 2013 00:16:26 UTC