httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Brian Behlendorf <br...@organic.com>
Subject Re: virtual hosts in 1.3
Date Tue, 26 Aug 1997 02:58:43 GMT
At 06:01 PM 8/22/97 -0700, Dean Gaudet wrote:
>Here is what 1.3 should be doing according to what I intended.
>
>Multiple-port support:  You should be able to take a set of multiport config
>files and split them into multiple single port config files just by changing
>the Listen statements, and nothing else.  From outside the box you shouldn't
>be able to tell if there's a single multi-port server, or multiple
single-port
>servers.  I think that succinctly describes the intention with ports, the
>rest of this message assumes we're dealing with a single port server.

Okay, I'll buy this.

>The main server is, well you know what it is.
>
>A name-based vhost is a <VirtualHost> section which had at least one ip
>address overlap the main server.  The other non-overlapping addresses are
>treated as if they're separate ip-vhosts with the exact same config.  A
>name-based vhost is matched only if the request is on a main server address,
>and ServerName, ServerAlias, or ServerPath match the name-vhost.  Matching
>occurs in the reverse order the hosts appear in the configuration files.
>Both ServerName and ServerAliases are considered at the same time, if
>there's no Host: header then ServerPath is considered.  For all intents
>and purposes any address strings appearing in the <VirtualHost> line
>are considered ServerAliases.
>
>An ip-based vhost is a <VirtualHost> section which does not overlap the
>main server's addresses.  An ip-based vhost is matched only by ip address,
>nothing else is considered.  A corollary is that no two ip-vhosts can
>(successfully) share the same ip address, the server does not presently
>diagnose this.

I guess I still don't understand why name-based vhosts can't be configured
on non-main addresses.  Alexei's example of two companies with ten
addresses each who want different responses to a Host:-less header makes
absolute sense, and was a real situation on Hyperreal (for
*.burningman.com).  The restrction just seems arbitrary.

	Brian


--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--
"Why not?" - TL           brian@organic.com - hyperreal.org - apache.org

Mime
View raw message