- From: Roy T. Fielding <fielding@gbiv.com>
- Date: Mon, 21 Sep 2015 15:04:27 -0700
- To: Stefan Eissing <stefan.eissing@greenbytes.de>
- Cc: Group HTTP Working <ietf-http-wg@w3.org>
> On Sep 21, 2015, at 5:06 AM, Stefan Eissing <stefan.eissing@greenbytes.de> wrote: > > Implemenation raised the question if tokens in Upgrade headers are to be treated case sensitive or not (for ALPN it is clearly defined). The current status > > - rfc7230 does not explicitly say so, but for other headers it does. So, one could say that Upgrade values are case-sensitive They are case-sensitive. > - rfc7230 reserves tokens of flavor "HTTP*" for itself and future versions > - the ALPN value "http/1.1" is seen in the wild ALPN has nothing to do with it. > - WebSocket RFC explicitly defines its token as case-insensitive And yet only registers one token. > I see only interop problems should HTTP insist on case sensitive values. Does 7230 need a fix here? No, interop problems will only occur when a string other than the registered token is sent. We should not encourage variations. ....Roy
Received on Monday, 21 September 2015 22:04:54 UTC