- From: Scott Lawrence <lawrence@agranat.com>
- Date: Fri, 25 Jul 1997 09:47:21 -0400
- To: Jeffrey Mogul <mogul@pa.dec.com>
- Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
JM> (8) [For the moment, two possible alternatives here!] I'm afraid that I can't decipher whether it is (8a) or (8b) that I like or whether I don't like either one. Simply stated, I would prefer that proxies never modify either Public or Allow headers; they should always be end-to-end headers. I choose this alternative because I think it is easier for a client to discover what is going on this way. If proxies DO modify the Public and Allow headers, an end-client can't tell whether it is the origin server or the proxy that doesn't support what it wants because it sees the same response from both. While it is true that the client may have another way to reach the origin server, if that is not the case there will be no way for the user to determine that the proxy is the problem. Since many users are behind proxies they cannot get around, there would be no way for them to discover that the reason they can't use some service is that the proxy is missing something. If proxies DO NOT modify Public and Allow headers, the methods supported by the origin server can be discovered by sending an OPTIONS request, and the methods supported by each proxy can be discovered by sending an OPTIONS request with a Max-Forwards header. This would be a slight change to RFC 2068 definition for the treatment of these headers by proxies, but as far as I can tell there are no deployed 2068 proxies out there yet, so I don't think that is a practical concern. Even if there were, a client could detect that a proxy was 2068-only by sending OPTIONS with a Compliance header to the proxy and get back an OPTIONS response without a Compliance header. -- Scott Lawrence EmWeb Embedded Server <lawrence@agranat.com> Agranat Systems, Inc. Engineering http://www.agranat.com/
Received on Friday, 25 July 1997 06:55:50 UTC