- From: <rlgray@raleigh.ibm.com>
- Date: Tue, 16 Sep 1997 12:42:26 EST
- To: HTTP Working Group <http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com>
Forwarding note from: Martin Presler-Marshall <mpresler@us.ibm.com> Tue, 16 Sep 1997 11:24:06 -0400 There is, I hope, one bug in this draft. In section 6, an example is given of a non-200 return code (specifically, a 407) from the proxy. After the plaintext HTTP header, the draft says that "...SSL data..." will be sent. This violates the requirement that the proxy need not have a complete SSL implementation. The proxy server, assuming that it doesn't actually understand SSL - or whatever protocol is being tunneled - can only sent back a standard HTTP/1.0 or HTTP/1.1 response. Therefore, the 407 response (or any other error response) should send back a content-type in its headers, and the appropriate type of response - unencrypted - in its content-body. This then begs the question of whether the "200 OK" response to a CONNECT request should include a content-type (our tunneling implementation does not currently send one). I think that it should, and a new MIME type like "application/tunnel" should be sent. This is obviously not a requirement, but would be nice for completeness. -- Martin Richard L. Gray chocolate - the One True food group cc: luotonen@netscape.com
Received on Tuesday, 16 September 1997 09:45:39 UTC