Path: utzoo!mnetor!uunet!munnari!moncskermit!goanna!yabbie!cit5!olympic!jim!jon
From: jon@jim.odr.oz (Jon Wells)
Newsgroups: comp.sys.nsc.32k
Subject: Re: NS32000 computer kit wishlist
Message-ID: <205@jim.odr.oz>
Date: 24 Feb 88 00:54:52 GMT
References: <3703@bcsaic.UUCP> <4953@nsc.nsc.com> <193@jim.odr.oz> <20411@amdcad.AMD.COM>
Organization: O'Dowd Research Pty. Ltd., Melbourne, Australia.
Lines: 72

In article <20411@amdcad.AMD.COM>, phil@amdcad.AMD.COM (Phil Ngai) writes:
> In article <193@jim.odr.oz> jon@jim.odr.oz (Jon Wells) writes:
> >Great but then what!!!
> 
> The vendor of the chip could let you use one of their systems which
> was licensed. And that system could be in your house or wherever you
> 
This article is perhaps the best response I received.
It is the best response only because it's the least likely!
All but one or two letters have basically said.......
"what's your problem, you can get a binary licence etc. etc."

  [the following is just a pet gripe skip it if you like]

  You must understand that this is the wrong side of the world to
  try such things. I doubt that you even need fingers to count the
  number of people at NS Aust. that have USED a unix system.
  Expecting them to have a machine AND the source to lend to anyone
  let alone a no one like me is just not on.

  It's not NS Aust's fault, their just a licenced sales office, as are
  most of the semiconductor people here, they have to pay for everything
  that they receive from NSC. 
  Unix to them is a great buzz word for the sales people to use and
  not much else.

  [What do you get if you rub two salesmen together,..TIRED ARMS!]

Now back to the real problem. A binary licence is great and cheap but
it's only of any use if the differences between the machines can be
hidden in the drivers. The second you try to do something different,
heaven forbid, you have no choice but to dive into the kernel.
Even just doing something `nice' with a graphics interface really
needs the support of kernel modifications.

It's not NSC's fault, it's not AT&T's fault, it's not UNIX's fault.
When the unix first grew up and started to get off the PDP's it was
hailed as a truely PORTABLE system.
Now we see computers advertized as "unix compatible", WHAT IS THAT???
I know! it's a machine that looks a lot like a Vax, architecturally anyway.

Try porting it to a machine that stores every data structure and function
in a seperate segment, where even trying to modify the base registers
that address these segments causes a protection fault and see how far you get.
I have some friends who ported a C compiler to such a machine,
the only (simple) solution was to give every C program a great big
segment all on it's own and let it do whatever it liked in there.

Enough drivel, the point that I am trying to make is that unix is
not a good thing for someone whose preferred pass time is designing 
strange hardware. It's just too big and complex. If the design effort
is to be worthwhile you have to be able to do something other than
run toy programs loaded from a cross-developement machine and that's
too much work for one person, even for a small group of people it
would take a very long time. Things that progress very slowly have
a tendency to never finish, you just spend your time changing details.
Maybe these sorts of projects don't have an absolute end, just getting
to the point of being useable is a good place to call finished.

I think the solution for me is a system written in a HIGH LEVEL language
that doesn't allow nasty's like *( (struct mess_up *) 0x42 ) etc...
A language which is far removed from any real machine (smalltalk?).

Maybe this is the wrong place to air my views but it is where I started.
It also seems to be the only vaguely hardware related group that is
received in Australia.

Yes I'm always this boring, jon.
-- 
Jon Wells @ O'Dowd Research P/L.  Ph: 03-562-0100 Fax: 03-562-0616

Rogue isn't the only place that leprechauns steal your gold!