W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > September to December 2000

Re: Proxy-connection

From: John C. Mallery <jcma@ai.mit.edu>
Date: Wed, 6 Sep 2000 11:24:43 -0400
Message-Id: <p04320407b5dc11d03b6c@[128.52.39.204]>
To: "Life is hard, and then you die" <ronald@innovation.ch>
Cc: http-wg@cuckoo.hpl.hp.com
By experimentation, I found that returning the usual
connection and keep-alive headers works for Netscape and IE.


At 14:22 -0700 09-02-2000, Life is hard, and then you die wrote:
>On Fri, Sep 01, 2000 at 10:19:32PM -0400, John C. Mallery wrote:
>> Can anybody give me a url to some documentation on this header,
>> and especially, how to get http 1.0 clients to do some rudimentary
>> persistence on requests through a 1.1 proxy?
>
>I don't know of any specs. But basically it works just like the ol'
>'Connection: keep-alive' header, except that it only applies to
>proxies. Of course it has the same basic problem as the original
>Connection header (see section 19.6.2 in rfc 2616) in that if you
>have two proxies in series and the first one doesn't understand
>the Proxy-Connection header and therefore forwards it to the second
>proxy which does understand it, you'll have a mess.
>
>Anyway, if you're writing a 1.1 proxy then you can certainly use the
>Proxy-Connection header on the client connection side (but only if the
>client is 1.0, of course), with the caveat that if that client is
>actually a 1.0 proxy which doesn't understand the Proxy-Connection then
>you're in trouble. I'm not sure if I understand your "how to get 1.0
>clients to ..." - other than sending back Proxy-Connection headers if
>the client sent one, and making sure a Content-Length header is present
>in the response as often as possible (this may mean generating it if
>possible), there's nothing you can do.
>
>
>  Cheers,
>
>  Ronald
Received on Wednesday, 6 September 2000 17:28:40 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:33:39 EDT