[net.micro.amiga] Unix, etc.

carroll@aim.rutgers.edu@caip.RUTGERS.EDU (02/17/86)

From: "Mark Carroll" <carroll@aim.rutgers.edu>

  I tend to agree with the earlier articles here. A Unix like system should
have been used as the primary OS for the Amiga. Yet another proprietary OS
is just a pain. I say Unix since it is a very powerful, and very standard
system. The Microworld is already too full of little OSs, each for one machine.
In the 68000 world, when there are already two Unix-like systems, one of them should be adopted. Modifying the Workbench to accept a Unix type OS would be 
a formidable proposition, but worth it for the benefits it would yeild.

  Are you listening Commodore?

      Mark Carroll
      Carroll@RU-AIM.Arpa
------

acs@amdahl.UUCP (Tony Sumrall) (02/20/86)

In article <1284@caip.RUTGERS.EDU> carroll@aim.rutgers.edu@caip.RUTGERS.EDU writes:

> From: "Mark Carroll" <carroll@aim.rutgers.edu>
> 
>   I tend to agree with the earlier articles here. A Unix like system should
> have been used as the primary OS for the Amiga. Yet another proprietary OS
> is just a pain. I say Unix since it is a very powerful, and very standard
> system. The Microworld is already too full of little OSs, each for one machine.
> In the 68000 world, when there are already two Unix-like systems, one of them should be adopted. Modifying the Workbench to accept a Unix type OS would be 
> a formidable proposition, but worth it for the benefits it would yeild.
> 
>   Are you listening Commodore?
> 
>       Mark Carroll
>       Carroll@RU-AIM.Arpa
Why does it *have* to be Unix?  From the discussions I've seen here quite
a few people (who *seem* to be in the know) don't really prefer Unix as a
model.  I've been using Unix for in excess of 4 years, MS-DOS and CP/M-80
for over 3 and, while I prefer Unix to the other 2, am not against someone
making another attempt at providing a "better" operating system/user
interface.  I must agree that AmigaDOS is not a panacea but I find it easy
to use and friendly enough.  Course this doesn't mean that you should stop
complaining--that's how things get improved.  Perhaps some pointed and
(more or less) precise suggestions should be proffered.
-- 
Tony Sumrall                       ...!{ihnp4,hplabs,amd,sun}!amdahl!acs

[ Opinions expressed herein are the author's and should not be construed
  to reflect the views of Amdahl Corp. ]

dillon@CORY.BERKELEY.EDU (Matt Dillon) (02/23/86)

	There has been quite a lot of talk about a UNIX or UNIX lookalike on
the amiga, some of it originating from myself.  Now that I have had a chance
to write some more indepth programs on my amiga, I would like to offer some
food for thought.

	Obviously, the Amiga's OS is setup for, well, the Amiga.  One reason
I think a UNIX like operating system would have been more benificial is that
we would then be able to port various utilities currently running on UNIX
systems over to the Amiga with relative ease.  UNIX hasn't been around for
countless years for nothing.

	But taking a good look at the Amiga's OS, I can find only two faults
with it:  The multi-processing was done incorrectly (e.g. if you run an lc1,
alink, or lc2 in the background, the rest of the machine freezes.  
Specifically, when the latter programs have finished their disk access and
are doing purely computational in-memory processing).  Secondly, the Amiga's
OS does not handle error conditions at all well, and doesn't have the 
ability to kill a process nicely.

	But then again, a UNIX without memory protection does is not 
guarenteed to solve these problems.  As far as a running system is 
concerned, the problems are trivial (that is, when you aren't attempting
to write a program).  Minor asside: Your Amiga is guarenteed not to stay
up for more than a couple of days (depending), due to it's habit of loosing
memory on an Execute() + other sources.  Another minor asside: The Amiga
has a memory-fragmentation problem, in that after using it for many hours,
Alink sometimes crashes the machine because it can't get a contiguous 
piece big enough.

	The Amiga's OS, being built especially for the Amiga, has the
freedom of being specialized.  Theoretically, it can handle it's
resources more efficiently than a mod'd UNIX.  The mod'd UNIX would have
to have another system layer in between, though this consequence reduces
system crashes due to error conditions by a considerable degree.

	Apart from the above mentioned problems, and discounted those other
problems which always appear in a new, innovative OS, I like the Amiga.
It certainly is much better than the MAC in terms of writing a program.

	The best idea, I think, is the 'shared libraries' concept.
In fact, I think Lattice made a *big* mistake by not making their basic
'system like calls' a library.  The equivalent UNIX thing would be yet another
system call.  Shared libaries are unrealistic in BSD UNIX's, at least until
the BSD gets between process shared memory working on a software level.

	If you think about it, we have been given unbelievable support in
terms of library routines for the Amiga.  The Graphics library, for instance,
is the most impressive and well written set of routines I have seen in a long
time.  By sticking so much support in, there are bound to be problems with
a few of them (Murphy's Law), though not all of them (Probability Law: The
more routines you have, at least one should work).  An example of that would
be the boched SER: driver, (and even the serial.device to some extent).

---------------------------------------------------------------

	Now what has everybody missed the most between UNIX and the Amiga?

	I'll tell you what we have all missed: A SHELL!.  

	aliases, history, filename expansion.....

	I thought you would be interested.


	I have written such a program.  It has nowhere near the power of
CSH, or even SH, and my lack of knowlege has prevented me from handling
Ctl-C, or return-error values, but it does have the following:

	aliases
	variables
	file-expansion ala  * & ? 
	history ala very simple (!pattern, !!, !number)

	it doesn't do everything correctly, but it certainly works better
than CLI in terms of doing programming from it.

	I'll post it as soon as I finish writing the manual for it.
If anything, you can get the independant file-expansion routines you've all
been craving for from it.  I don't purport that it's complete (it isn't), or
completely error free (it isn't) but.... well see.  

				-Matt

james@uw-june.UUCP (02/26/86)

In article <8602230657.AA09363@cory>, dillon@CORY.BERKELEY.EDU (Matt Dillon) writes:
> 	But taking a good look at the Amiga's OS, I can find only two faults
> with it:  The multi-processing was done incorrectly (e.g. if you run an lc1,
> alink, or lc2 in the background, the rest of the machine freezes.  
> Specifically, when the latter programs have finished their disk access and
> are doing purely computational in-memory processing).  Secondly, the Amiga's
> OS does not handle error conditions at all well, and doesn't have the 
> ability to kill a process nicely.
  I'm not convinced that AmigaDos should get all the blame for these
problems: The compiler and linker (especially the linker) seem to be poor
examples of how to write a program to fit in to this multi-tasking
enviroment. It should be possible for programs to written which keep good
track of their resources (using Remember?) and can be deleted.
  In large part these problems seem to stem from the fact that there is no
MMU to aid the CPU. If different tasks ran in different memory spaces, or a
2 dimensional space like Multics or VMS, then reclaiming the resources would
be greatly simplified.
> Another minor asside: The Amiga
> has a memory-fragmentation problem, in that after using it for many hours,
> Alink sometimes crashes the machine because it can't get a contiguous 
> piece big enough.

  Granted: Alink is very unimpressive.

> 	Now what has everybody missed the most between UNIX and the Amiga?
> 	I'll tell you what we have all missed: A SHELL!.  
> 	aliases, history, filename expansion.....
> 	I thought you would be interested.
  I'm very interested Matt. Sounds just great. Does your shell handle path
searching (as in unix with the PATH variable)?

- James Synge