W3C home > Mailing lists > Public > ietf-http-wg@w3.org > April to June 2003

Re:Re: Proxy and HTTP protocol versions

From: Sudha <ssudhaiyer@hotmail.com>
Date: Sat, 7 Jun 2003 22:29:01 -0400 (EDT)
To: <ietf-http-wg@w3.org>
Message-ID: <Law14-OE14ETiTSUPDc0003a4f0@hotmail.com>

Firstoff, thanks for all the response. I've provided my answers
with some more questions inline (enclose between >>>>)

-Thanks/Sub

There are a few questions you need to answer before we can
help you with the design and minimal feature set:

	1. Will your proxy forward all requests to one configured
	   server? Or will it need to look at requested URLs and
	   forward requests to corresponding servers?

>>>>>>>>>>>>
Yes. My proxy server would always forward it to one configured server. I don't have
to lookup the request header. 
>>>>>>>>>>>>


 2. Is performance important? Do you want your proxy to cache
	   responses?

>>>>>>>>>>>>
No. I don't have to provide any caching feature.
>>>>>>>>>>>>

 3. Will you proxy do anything other than forwarding messages?
	   That is, why do you need a proxy?

>>>>>>>>>>>>
The reason I need this proxy is because, I want a way to forward all requests
to a centralized server. And I want to write one on my own. 

The end server that I would be sending the request to is HTTP 1.1 compliant. But,
I doubt if the server properly implements 'Keep-alive' etc. 

My question is, if there din't exist a proxy between the browser and the server,
won't the browser and the server handle all these differences ? In which case,
how should a proxy now make a difference ? Can't I just forward the request
or just tunnel it as if it were HTTPS, and be assured that it works?

>>>>>>>>>>>>

Your answers will determine how much HTTP corners you can cut. If it
turns out that your proxy needs to support HTTP/1.1, then you would
have to handle versions and *Connection headers correctly, among other
things. In that case, I would not recommend just removing
Proxy-Connection; a lot more work needs to be done to survive in a
general HTTP environment (RFC 2616 defines most of what needs to be
implemented in that case).

Alex.
Received on Sunday, 8 June 2003 05:31:34 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 06:49:23 GMT