[Coco] TCP/IP Programming in Commodore BASIC

[Aaron Wolfe]

On Sat, Nov 6, 2010 at 11:27 AM, Gene Heskett <gene.heskett at gmail.com> wrote:
> On Saturday, November 06, 2010 11:18:06 am Aaron Wolfe did opine:
>
>> I'd like to see an IP stack on the CoCo as well.  There has been
>> discussion here about such a thing in the past.  Maybe it would be a
>> good topic for the next coco conference.
>>
>> Ideally an IP stack would be accessible from both DECB and OS9.  I've
>> looked at the cheap (< $5) "TCP/IP on a chip" solutions that are
>> available and thought that one of these would make it easier to add IP
>> to the CoCo, since you would interface to an existing stack rather
>> than creating your own.  These chips would buffer and decode the IP
>> packets into data streams and also have the interface used to talk
>> with ethernet or wireless networking hardware built in.  They would
>> let the CoCo use it's resources at the application layer rather than
>> dealing with the protocols.
>
> Didn't the majority of those require far more than the usual 4 address for
> a full interface to them?  Like about 16 of them?  That right there would
> seem to bloat the driver considerably.

see below...

>
>> The last time I mentioned such a thing it wasn't a popular idea.
>
> You didn't hear negativity that from me.
>
>> It
>> was suggested that if we don't run the IP stack on the 6809, we
>> shouldn't do it.   I'd be happy with that approach if we had a stack
>> for the 6809, but as far as I know no one is working on one.
>
> KA9Q could do slip only, not tcp/ip.  & that was crashy as could be.  And
> by that time, no ISP's were offering slip connections in my neck of the
> woods anyway.
>

I could add slip support to DriveWire.  You could then use one (or
more) of the dw virtual channels as a slip device, they just look like
regular serial ports to OS9.  I don't think this is an awesome
solution and it wouldn't bring IP to DECB, but if anyone is interested
I'm willing to give it a shot.  I'd guess it could be made to work
with KA9Q, but sounds like that might not be ideal.

>> Writing a stack isn't interesting to me, although using one is :)
>> That's one reason why the IP services provided by DriveWire are
>> implemented as regular serial channels on the CoCo side.. I just don't
>> care too much how IP gets there so long as it's there.
>>
>> I think it would be valuable to have a real IP stack no matter how
>> it's provided, and I'd be willing to help with an effort to implement
>> any sort of IP solution for the CoCo.  I don't think there is much
>> point in extending the IP functionality in DriveWire unless it's to be
>> compatible with some other "real", coco native solution.
>
> I am still in favor of the 'chip with a stack' solution, if some way to
> squeeze its register addressing into the coco's normal I/O map could be
> done.  However, I have not investigated today's offerings in that area.
>

A few months back I had some interesting conversations with Walter
(believe he is still on the list, maybe could clarify this) about
interfacing between the CoCo and a TCP/IP controller.  I am by no
means a hardware guy and don't understand the details, but I believe
the idea was that a 6551 could be the interface between the 6809 and
the controller.