[Coco] os9/drivewire driver: success!

[Aaron Wolfe]

On Sun, Nov 8, 2009 at 8:54 PM, Boisy G. Pitre <boisy at tee-boy.com> wrote:
> On Nov 8, 2009, at 7:46 PM, Aaron Wolfe wrote:
>
>> I am super pleased to report, despite many remaining issues, I now
>> have a working terminal over the drivewire.  So a single cable is
>> proving disk, clock, print *and* a terminal on T2!  Drivewire is so
>> nice.  Now I need to flesh out the driver, its very incomplete, but
>> the concept works.  I am so excited :)
>>
>> I had things sort of working all along, I just didn't understand that
>> Read is supposed to sleep until it can get a byte.  I was returning
>> when I had nothing, causing the problems I mentioned in earlier posts.
>
> Of course... V.BUSY/V.WAKE.  It's amazing what you forget when you haven't
> stuck your head into something for a while.
>
>> Here is an example of things working now (note process 3, this is the
>> [...]
>> OS9:
>
> You rock, man!  Great work!
>
> What's your next step?

Well, first I've got to improve the driver to actually handle errors
and be more efficient (flush entire buffer instead of character by
character, etc).

The long term goal is twofold: first to implement a useful full screen
terminal with coco control codes translated into curses operations.
second, to make a "modem" in the server, this is what really interests
me.  the "modem" will respond to standard hayes commands, but connect
to IP sockets.  So, an OS9 terminal program might send
ATD192.168.1.1:4000, the server will connect to 192.168.1.1 via TCP,
send CONNECT to the Coco, and then connect input/output.   For
incoming connections, the server will listen on a port and send "RING"
or maybe set RI high, then connect the incoming tcp connection.

Ultimately, this will allow one Coco to dial another Coco (BBS, etc),
over the internet :)