From: Chuck Murcko <chuck@telebase.com>
Message-Id: <199607162027.QAA09201@telebase.com.>
Subject: Re: Broken clients
To: new-httpd@hyperreal.com
Date: Tue, 16 Jul 1996 16:27:38 -0400 (EDT)
In-Reply-To: <9607161938.aa27693@gonzo.ben.algroup.co.uk> from "Ben Laurie" at
 Jul 16, 96 07:38:07 pm
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-new-httpd@apache.org
Precedence: bulk
Reply-To: new-httpd@hyperreal.com

Ben Laurie liltingly intones:
> 
> OK, it seems we now have two reasons to detect broken clients. One is to avoid
> flushing after the headers, the other to disable keepalive. Does anyone have
> a recipe for detecting those clients?
> 
The following code at the beginning of set_keepalive() should take care of
that part:

int set_keepalive(request_rec *r)
{
  char *conn = table_get (r->headers_in, "Connection");
  char *length = table_get (r->headers_out, "Content-length");
  char *agent;
     
  agent = table_get(r->headers_in, "User-Agent");   
  if(strncmp(agent, "Mozilla/2", 9) == 0)
    return 0;

  ...

Same could go into send_http_header()

    agent = table_get(r->headers_in, "User-Agent");
    if (c->keepalive && strncmp(agent, "Mozilla/2", 9) == 0)
        bflush(r->connection->client);  /* For bugs in Netscape, perhaps */

unless we wanted a check_for_bogus_mozilla() added to http_protocol.c
for this purpose...

chuck
Chuck Murcko	N2K Inc.	Wayne PA	chuck@telebase.com
And now, on a lighter note:
Lie, n.:
	A very poor substitute for the truth, but the only one
discovered to date.