W3C home > Mailing lists > Public > www-html@w3.org > March 2003

"403 Found", old browsers, and name-based virtual hosting

From: Erica Douglass <erica@simpli.biz>
Date: Fri, 14 Mar 2003 10:13:19 -0800
To: <www-html@w3.org>
Message-ID: <008801c2ea55$64697db0$ac7ba8c0@corkyserver>

Hi all,

I was surfing the web a bit in Mosaic today for nostalgic purposes, and
I noticed something curious. While sites like Google and Yahoo loaded
decently, any sites I went to that were hosted on our servers made the
browser fall into an infinite loop.

To see why, I tailed our browser logs and grabbed the command that
Mosaic was using -- a simple GET / HTTP/1.0. I telnetted in to a couple
of the domains we host on port 80 and issued the same command. I got
back an odd response -- a 403 Found with a link to the same website I
had just requested. For instance, telnetting to www.onthehouse.com port
80 sends the Found URL back as http://www.onthehouse.com. No wonder this
poor old browser was falling into an infinite loop!

I vaguely remember that this has something to do with name-based virtual
hosting. I have two questions. 

1. What do modern browsers do to get around this? I understand that they
send some sort of HOST header, but what should I do e.g. if I am
telnetting into the server?
2. Would turning off name-based virtual hosting for sites that are using
their own IP address result in a performance gain? The browser would not
have to send two requests... but perhaps modern browsers always send the
HOST header and thus name-based virtual hosting is no slower than
regular IP-based hosting.

These questions are more curiosity than anything, as I continue to learn
how HTTP works and what modern browsers do that older ones didn't.
Thanks in advance for an explanation!

Erica Douglass
Lead Web Developer
Simpli, Inc.
Received on Friday, 14 March 2003 13:14:56 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:06:03 UTC