Path: utzoo!attcan!uunet!wuarchive!usc!ucsd!helios.ee.lbl.gov!nosc!crash!pnet01!jca
From: jca@pnet01.cts.com (John C. Archambeau)
Newsgroups: comp.os.minix
Subject: Re: Unix for Atari 8-bit
Message-ID: <2870@crash.cts.com>
Date: 29 May 90 01:16:02 GMT
Sender: root@crash.cts.com
Organization: People-Net [pnet01], El Cajon CA
Lines: 58
X-Local-Date: 28 May 90 18:16:02 PDT

MAB01057%UFRJ.BITNET@cornellc.cit.cornell.edu (Marcelo Amarante Ferreira Gomes) writes:
>I missed this stuff while our link to the net was down, but, as an
>Atari 8-bit user (I still use it as a terminal for my PC), I have some-
>thing to say: It's a dream I'd like to see, but I doubt it'll be
>practical or even pleasant to use, for it's disk drive is sooooooooooooo
>slooooooooooooooow... I think that on any machine wich communicates to
>its drive via a serial line, Unix, Minix or *ix will never feel the way
>it really is.
>
>But if you face it as a programming challenge, though, it might be well
>worth to see if one can write such a piece of code - I'd face it myself
>if I weren't so busy working on a compiler (during the free hours) with
>Guy Helmer. Anyway, the request is pending on my (rather long) future
>projects list. Maybe I can later fix the compiler I'm writting to cross
>compile code for the 6502.
>
>I've already given it a little thought: we could simulate a larger stack
>using a 2-byte stack pointer on page zero - that wouldn't be so slow,
>since page zero access is so fast. Regarding the OS itself, one could
>use the display list interrupt as the time-slice end marker.
>
>Well, let's stop here, for old 8-bit Atari fans like me may get nostalgic
>and start writting the kernel from a series of articles starting with
>this one. If anyone has suggestions (or would like them), you may get in
>touch with me and maybe something comes out. Please don't expect me to
>get too serious about this. I have a college graduating course to finish
>and I think I'm already much busyer than I can handle, but I sure appreciate
>a nice chat :-)

One serious problem I forsee (and I don't even know if you can get around it)
is how the Atari SIO bus handles devices.  If you have a serial port open, you
have to have the channel to your disk drive closed.  This is how all terminal
and BBS programs handle disk I/O.  So during a sync, every thirty or so
seconds the user would experience this slow down because of:

Close RS232 so we can yank the bus for disk I/O
Do sync
Reopen RS232 port

And this would go on every time a disk I/O would be done.  This is assuming
that you're going to make your Atari 8-bit Minix as Unix like as possible.

Whether ICD's MIO board behaves this way or not, I don't know.  I do know that
if you have a storage device daisy chained on the SIO port, you have to do
this because of the small number of control lines.  And one of them being one
IRQ which is enough to discourage me from attempting Minix on an Atari 8-bit. 
At least with an IBM clone, you have some IRQ's to play with.
 
     // JCA

 /*
 **--------------------------------------------------------------------------*
 ** Flames  : /dev/null                     | Small memory model only for
 ** ARPANET : crash!pnet01!jca@nosc.mil     | Unix?  Get the (*bleep*) out
 ** INTERNET: jca@pnet01.cts.com            | of here!
 ** UUCP    : {nosc ucsd hplabs!hd-sdd}!crash!pnet01!jca
 **--------------------------------------------------------------------------*
 */