- From: William A. Rowe Jr. <wrowe@rowe-clan.net>
- Date: Tue, 26 Oct 2010 23:29:33 -0500
- To: Willy Tarreau <w@1wt.eu>
- CC: Mark Nottingham <mnot@mnot.net>, Julian Reschke <julian.reschke@gmx.de>, Adrien de Croy <adrien@qbik.com>, HTTP Working Group <ietf-http-wg@w3.org>
On 10/26/2010 10:52 PM, Willy Tarreau wrote: > On Wed, Oct 27, 2010 at 12:58:05PM +1100, Mark Nottingham wrote: >> I think we can specify: >> >> 1) CONNECT requests MUST have a zero-length body (same language referring to p1 as we used for 205) >> 2) CONNECT responses that are successful (2xx) MUST have a zero-length body, because the tunnel begins after the header block. >> >> Thoughts? > > I think this is fine. Concerning the last point, I think we could still > improve it a bit. By default, if there is no C-L, a user agent (or a > gateway) may consider that the body runs till the close (as it does > with other methods or statuses). As it is here, it makes one think that > it is mandatory to send a content-length: 0. Maybe we should word it > slightly differently, by first indicating that the tunnel begins after > the header block, then that the receiver of the response must ignore any > body in such a case. Maybe something around this : > > 2) Successful CONNECT responses (2xx) indicate that the tunnel begins > immediately after the header block, regardless of any Content-Length or > Transfer-Encoding headers, which MUST be ignored by the recipient. These > responses MUST have a zero-length body and MUST NOT be transfer-encoded. > > Any thoughts ? The closer we can get to the language in 10.1.2 101 Switching Protocols the more likely this message will be clearly understood.
Received on Wednesday, 27 October 2010 04:30:25 UTC