Frank A. Vorstenbosch wrote:

> But beware!  Ports that are 'virtualized' under NetWare DOS clients
> have "NW" as their address value.  This is done so that programs that
> test just for presence of the port by looking for zeros find a port
> there (and writing to the port by means of the BIOS then will work
> because the NW client traps that).

  An interesting detail which I will spirit away.  Not terribly
practical though.  Prior to Windoze, no usable BIOS *existed* for the
COM ports and therefore all comms programs contained their own COM port
drivers, or used generic ones in which case NetWare *may* have provided
network-compliant versions of some popular generic drivers.

  Windoze contains comms drivers, but if you try to access COM ports
directly, it gets a bit upset, particularly if it determines a program
(and that includes the O/S itself, and notably the mouse driver) is
already using that port.

  In summary, those that *knew* pointed out that you can "bit-bang" the
TX Data line quite effectively using the "Set Break" bit, though I
wouldn't guarantee accurate timing.  Bit-banging the RTS and DTR lines
is also very easy and routinely used to implement synchronous protocols,
receiving data on the CTS, DTR and CD lines in the same fashion.  This
can even be done while the TXD is in use for normal, separate serial
data (just as LapLink does!).  *All* these lines are RS-232 (insofar as
the machine provides) at the connector.

  OTOH, most relays would not respond to the sub-millisecond
interruptions of a continuous stream of $FF characters sent at 1200
baud and above anyway.  Indeed, capacitive coupling and a rectifier
would make a very good "fail-safe" system with such a signal source.


Cheers,
        Paul B.