- From: Dan Winship <dan.winship@gmail.com>
- Date: Mon, 26 Nov 2007 15:56:41 -0500
- To: Adrien de Croy <adrien@qbik.com>
- CC: ietf-http-wg@w3.org
Adrien de Croy wrote: > Actually there are (I'm fairly sure) cases where a 200 response comes > back to a CONNECT request with an HTML body. > > For instances proxies that return a challenge page for authentication. That won't work. (Just tried it with Firefox.) I think you're thinking of the case where the proxy returns a 3xx response, to *redirect* the client to a login page. > Maybe the requirement should be that there must be a Content-Length set > (regardless of whether there is a non-zero length body), or other more > normal methods of message delineation before entering the tunneling phase. That would be a change to the way CONNECT works though. My suggestion was just updating the spec to match how things already work. -- Dan
Received on Monday, 26 November 2007 20:56:52 UTC