Date: Fri, 28 Feb 1997 17:04:56 -0500
Message-Id: <97022817045675@decus.org>
From: coar@decus.org (Rodent of Unusual Size)
To: New-HTTPd@hyperreal.com, Coar@topaz.decus.org
Subject: Re: [PATCH] PR#39 (no "Vary: Host" in response)
Sender: new-httpd-owner@apache.org
Precedence: bulk
Reply-To: new-httpd@hyperreal.com

>From the fingers of Alexei Kosut flowed the following:
>
>Sorry, that's not how Apache works. Apache looks at the Host header in
>*every* request. Any request can be transferred to any other server in
>the configuration by the use of the Host header.

    I see; that aspect escaped me.  Well, I said it was unfamiliar
    territory.  Live and learn.

>Roy (who, if you may recall, wrote much the HTTP/1.1 spec) has assured
>us that it is not neccessary to include Vary: Host, as this should be
>assumed by HTTP/1.1 proxies.

    Curiously enough, PR#39 was entered by (or is at least attributed
    to) Roy himself, in December 1996.  That's partly why I included
    "Roy..?" in the original patch submission.  Looking at the
    discussions of this issue since then, I don't see Roy saying
    anything resembling the first part, although the second part is true
    enough.  He says:

>                               .. the basic discrimination is whether the
>same request with a different Host value would result in a different
>resource being selected by the server.  If a particular server does not
>look at the Host value for requests on a particular IP address, then it
>does not need to send Vary: Host.

    and that's almost his last word on the subject.  Of course, I have
    probably missed something.

>Y'know, we've had this discussion at least twice in the last few
>months. There is a mail archive (http://dev.apache.org/mail/) and it
>seems to me that if you are going to bring up a topic on the list, you
>should check to see if it's been discussed recently first. It would
>save us all a lot of effort.

    It would save even more effort if bogus problem reports weren't left
    open.  I brought up the topic to try and close something listed as a
    bug, not out of a clear blue sky to waste everyone's time.  So,
    since this appears to have at least two -1s already, I'll withdraw
    the patch and close the PR to avoid anyone else running afoul of
    this.  If it's valid, it can always be re-opened.  If it's truly
    dead, this should be the last message (from me, anyway <g>) on the
    issue.

    #ken    :-/}