W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > September to December 1995

Re: Decision about Host?

From: Shel Kaphan <sjk@amazon.com>
Date: Sat, 7 Oct 1995 16:29:07 -0700
Message-Id: <199510072329.QAA13668@bert.amazon.com>
To: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
William Perry writes:
 >   The spry server can listen to more than one port (so that you can have
 > SSL and HTTP/SHTTP within one incarnation of the server, sharing the same
 > pool of proceses, etc).  I plan on extending this so that you can have
 > multiple configs for different interfaces all from one incarnation of the
 > server.
 > 

That's a nice touch.

The thing that worries me about expecting a "server proxy" (or reverse
proxy, or whatever we call it) running on a firewall or other gateway
to sort out all the port number issues is that in a large net, the
administration of the firewall (and the HTTP proxy running on it, if
any) may be fairly independent of the administration of any HTTP
servers running inside the network.  The site administrator may just
say that requests on ports N through M will be accepted, but will
all be forwarded on port P.  (Some sysadmins can be uncooperative, or
overworked).  Then the question of how to detect which port a request
is on is deferred to the origin server.  So, in this case, unless
there were a way for the port number to be a part of the request,
there would be no way for a server listening at a certain IP address
to disambiguate.

Since it seems a little unnatural to have the port number as part of
the Host header, what about an optional "Port" header that would
allow, for instance, proxies to pass information about the original
request port, without having to forward the request on a given port or
modify the request-URI?

Shel Kaphan
Received on Saturday, 7 October 1995 16:35:08 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:31:33 EDT