NetNews Usenet Archive 1992 #18/NN_1992_18.iso/spool/comp/sys/amiga/programm/12296

Text File | 1992-08-13 | 3.5 KB | 63 lines

Newsgroups: comp.sys.amiga.programmer
Path: sparky!uunet!usc!sdd.hp.com!hp-cv!hp-pcd!hpmcaa!hpmcaa!bryanf
From: bryanf@hpmcaa (Bryan Ford)
Subject: Speculations: Amiga and Mach?
Message-ID: <[e-mail]>
Sender: [e-mail] (News Administrator)
Nntp-Posting-Host: hpmcaa.mcm.hp.com
Organization: HP Cardiology Business Unit - McMinnville, OR
X-Newsreader: Tin 1.1 PL4
Date: Fri, 14 Aug 1992 01:33:28 GMT
Lines: 50

Lately I've been sifting through reams of PostScript documentation on the
Mach 3.0 microkernel, exploring its source code, etc.  I've also been hearing
quite a bit about it lately in other circles.  The mythical GNU Hurd, which
is likely to be extremely popular among the Amiga's hackers when released, is
Mach-based.  IBM's OS/2, which is already quite popular on PC's is going to
be moved to Mach and (from what I hear) become semi-portable in the future.
There are already several versions of Un*x running on various versions of
Mach, including the NeXT.  It seems to be getting popular, to say the least.

I don't want to cause an OS-war ruckus here, but most of us agree that one
major sore spot of the Amiga's OS is its lack of virtual and protected
memory.  (Anyone who's about to say, "but we don't *need* it", please think
back to our nice discussions about multitasking with the Atari ST chaps...)

Of course, it's obvious where I'm heading: have Commodore's engineers
considered migrating the Amiga's OS to Mach in the future?  If not, it may
be something to think about.

Such a switch would probably break zillions of programs, but because of many
parts of the Amiga OS's current design, that would be true of *any* move to
protected/virtual memory.  The choice is this in any case: either break lots
of existing software, or let the Amiga's OS become very outdated in this area.
And the breakage problem probably wouldn't be all that bad anyway:  I think
compatibility with a large number of existing programs could still be achieved.

So assuming Commodore is planning on moving to virtual/protected memory
anyway, it seems like using Mach 3 could be quite beneficial.  It is already
working well, and it will likely become very mature and reliable by the time
"Kickstart 3.0" could be released for it.  It would save Commodore all the
hassles of coming up with their own virtual/protected memory kernel design
and making it fast and reliable; instead they could spend their time making
the Amiga's OS work better with it.  All the Un*x hackers that have
traditionally gravitated toward the Amiga because of its similarity to Un*x
would be especially happy to be able to run the Amiga's OS and the Hurd or
other Mach-based operating systems alongside each other.  The Mach design
is clearly a design for the future, and to top it off, it's free.

Of course, the decision will be commodore's, so why am I not sending this
off to "suggestions@cbmvax..." instead of posting it here?  I want to see how
much other interest there is among Amiga programmers, discuss the potential
gains and problems, try to find the answer to the million-dollar question
"Just how much software would have to be broken?", and in general, just make
some noise.  Lots of "hot air" off of UseNet has made its way into Kickstart
2.0 in the form of some really great features.

If many people aren't too familiar with Mach, say so, and I'll write up and
post a brief introduction.  (People who are *really* interested can do like
I did - download the PostScript docs from mach.cs.cmu.edu.)

                Bryan Ford