To: new-httpd@hyperreal.com
Subject: Re: more Keep-Alive - possible problem 
In-reply-to: Your message of "Tue, 06 Feb 1996 23:12:05 PST."
             <Pine.HPP.3.91.960206225713.5582A-100000@ace.nueva.pvt.k12.ca.us>
Date: Wed, 07 Feb 1996 13:08:33 -0800
From: "Roy T. Fielding" <fielding@avron.ICS.UCI.EDU>
Message-ID: <9602071308.aa09159@paris.ics.uci.edu>
Sender: owner-new-httpd@apache.org
Precedence: bulk
Reply-To: new-httpd@hyperreal.com

> At any rate, I've further localized the bug. It's not, as I reported 
> earlier, 67 byte files and smaller, but rather HTTP responses smaller 
> than 255 bytes (including both the response-line, headers and body). 
> Which at least is a more round number (in the binary sense). It also means 
> we could stick in a bogus header that would round out the number of bytes 
> to >255...

Here ya go....

    Warning: 99 Your browser sucks.  This really long message is here because
             someone didn't bother to test their implementation of Keep-Alive.
             Maybe you should think about upgrading it.

> Besides, don't knock it 'till you've tried it. I've confirmed that this
> bug exists in Netscape/Mac, on several systems, and I believe Netscape
> uses shared code across their platforms for this sort of thing, and I was
> wondering if anyone has attempted to see if this exists on other OSes,
> which I don't have access to. So if someone could point a Unix or Windows
> version of Netscape 2.0 at http://www.nueva.pvt.k12.ca.us:8001/, and tell
> me what happens (it's a 10 byte file, which makes 195 bytes w/ headers,
> and causes the Mac version to hang for 15 seconds at "100%" until the
> connection times out), that'd be helpful. I haven't yet filled out a bug
> report form at Netscape, I'd like to get all my facts straight first.

The bug is present in 2.0b4 on X/SunOS.

.....Roy