From: Jim Jagielski <jim@jaguNET.com>
Message-Id: <199608231339.JAA03022@shado.jaguNET.com>
Subject: Re: nph and lingering_close
To: new-httpd@hyperreal.com
Date: Fri, 23 Aug 1996 09:39:02 -0400 (EDT)
In-Reply-To: <Pine.SOL.3.93.960822150607.18154A-100000@acidik.organic.com>
 from "Alexei Kosut" at Aug 22, 96 03:12:19 pm
Content-Type: text
Sender: owner-new-httpd@apache.org
Precedence: bulk
Reply-To: new-httpd@hyperreal.com

Alexei Kosut wrote:
> 
> Okay...
> 
> I've spent the last few hours trying to figure out why nph scripts don't
> work with the lingering close code. It boils down to something like this:
> 
> When mod_cgi starts an nph script, it duplicates the client connection fd,
> and passes it to the CGI script as stdout. Because there are two handles
> to this fd, when bclose() closes its, it doesn't actually close the
> socket because the CGI script still has one open. The connection stays
> open until the CGI script exits. However, lingering_close() uses
> shutdown(), which closes the socket right away, even though the CGI script
> still has a handle open to it.
> 
> So... the fix? Well, a simple one is to simply make mod_cgi wait until the
> CGI script finishes before passing control back to Apache. This way you
> sidestep the problem entirely. Patch follows:
> 

Nice and elegant. Works like a charm too!

+1
-- 
Jim Jagielski  << jim@jaguNET.com >>   |   "There is a time for laughing,
  **  jaguNET Access Services  **      |    and a time for not laughing,
      Email: info@jaguNET.com          |    and this is not one of them"
++    http://www.jaguNET.com/         +++      Voice/Fax: 410-931-3157       ++