X-Authentication-Warning: delorie.com: mail set sender to opendos-bounces using -f
Message-Id: <200501281919.j0SJJp6H000385@delorie.com>
From: "Michal H. Tyc" <mht AT bttr-software DOT de>
Organization: BTTR Software
Date: Fri, 28 Jan 2005 20:06:05 +0100
X-Mailer: Arachne V1.75+/3C+
To: opendos AT delorie DOT com
Subject: Re: PNW vs NetWare (and NE2000 drivers, Client32)
MIME-Version: 1.0
Content-type: text/plain; charset=iso-8859-2
Content-transfer-encoding: 8bit
Reply-To: opendos AT delorie DOT com

On Thu, 13 Jan 2005 10:31:54 +1100, da Silva, Joe wrote:

> As for Client32, this is the 32-bit DOS (and MSW, I think)
[...]
> If this client were
> compatible with PNW (my interpretation of Ben's posting
> is that it isn't, only the 16 bit VLM client is), then I would
> definitely use this, because it of the major benefits it gives
> for DOS memory availability.

No, it's compatible with NetWare 3+ (or 4+?) servers only and not
with PNW, which uses different protocol.

> So, it seems there is a bug (actually, two bugs, as you
> will see) with the VLM ODI NE2000 driver. It performs
> some sort of test to check if the network is correctly
> terminated, but then misinterprets this as a hardware
> configuration conflict. Furthermore (and this is the second
> bug I alluded to), it then neglects to restore the interrupt
> vector for the card's IRQ, even though it isn't loading. As
> a result, attaching a cable or terminating resistor to the
> card's BNC connector will subsequently invoke a call to
> this interrupt vector, resulting in a hung PC - nasty!

Interesting... I'm using an NE2000 clone (Genius IIRC) in one
machine and it was happening from time to time that the cabling
was broken somewhere outside my room, or that I had to disconnect
the cable to use it with another machine. The driver loaded and
I didn't experience any problems of this kind, just observed
a noticeable delay during startup.

Regards,

Michal