Path: utzoo!mnetor!uunet!seismo!sundc!pitstop!sun!amdcad!ames!hao!oddjob!uwvax!umn-d-ub!umn-cs!meccts!nis!stag!trb
From: trb@stag.UUCP ( Todd Burkey )
Newsgroups: comp.sys.amiga
Subject: Re: MIDI/GAMES/etc (was Re: Serial)
Message-ID: <343@stag.UUCP>
Date: 18 Feb 88 04:38:47 GMT
References: <2287@crash.cts.com> <945@polyslo.UUCP> <332@splut.UUCP> <318@stag.UUCP> <1452@sugar.UUCP>
Reply-To: trb@stag.UUCP ( Todd Burkey )
Organization: Mindtools ST Access Group, Plymouth, MN
Lines: 40
Keywords: Serial, null modem, rs-232

In article <1452@sugar.UUCP> karl@sugar.UUCP (Karl Lehenbauer) writes:

>On commercial 
>token ring implementations, such as Apollo's, there is a relay on each 
>workstation's network board that causes the machine to be electrically 
>disconnected from the net when the machine is down or when it's up but 
>the network software isn't running.  (Granted, there are pathological
>conditions that can still break a net)

We have around 300 Apollo's on two rings at work...it is really easy
to break a net :-(. Remember also that that relay wasn't always on the
Apollo boards. We got along fine during the early Apollo workstation
years without such a switch.  Then networks got so large we started demanding
such a beast.

>On a machine implemented the way 
>the ST is; that is, a machine that needs to be rebooted a lot in the
>normal course of it's operation, this is fatal for anything but hoyybist
>get-togethers.

Fortunately it is mostly the hobbyists that spend all their time
rebooting their computers.  I've used ST's in a normal business environment
(engineering and software development) for days on end with out seeing
a crash. As a developer, however, I think I can safely say that
neither the Amiga or the ST (I used to have both) could make it
through a night of programming without resetting and maybe even
rebooting a couple of times (my programming anyway, since I always
seem to save up my goofups for when I am in in supervisor mode...)

If properly implemented, I would imagine that a Midi Ring could even
handle the occassional reboot without any glitches. Midi-Maze seems to
handle this pretty well...you can remove a machine from the network
without causing everyone to reboot, but obviously the net is down
until the connection is re-established. I can envision a little
heartbeat controlled bypass circuit that could be built for several
dollars per machine, but it just wouldn't be worth the effort on small
networks...

  -Todd Burkey
   trb@stag.UUCP