Re: [9fans] missing compilers?

Re: [9fans] missing compilers?

[David Gordon Hogan]

> I'm told that in
> the past Apple operated on the same principle as NeXt: that details of
> the hardware such as ROM contents could be changed on a whim since the
> company could then just tweak the OS to match.  Thus the hardware was
> effectively undocumented, or at least underdocumented, from the point
> of view of someone writing code for the bare hardware.

As I recall, the Mac Plus was pretty well-documented, but after
that, good hardware documentation became increasingly sparse.

The (``humble'') PC, on the other hand...

Re: [9fans] missing compilers?

[Douglas A. Gwyn]

jkw@bowser.eecs.harvard.edu wrote:
> In article <3D4F8DF3.18613.27913A3F@localhost>, Andrew Simmons wrote:
> > bolted on top of a "new" operating system which is based on
> > Carnegie-Mellon's 1980's version of an operating system from the
> > 70's.  At which point you start to think that even Bill Gates never legged
> > you over to quite this extent, and start looking at Windows machines.
> Ah.  I see, you're from the VMS camp, then?

It's even funnier.  OpenVMS evolved from VAX/VMS which was largely
inspired by RSX-11M, which evolved from RSX-11D which was a follow-on
to DOS/BATCH-11.  Windows XP evolved from a combination of Windows/NT
(which was largely inspired by VMS) and 16-bit Windows, which evolved
in several stages from a bolt-on to MS-DOS.  Forward compatibility
has always been a factor; even the latest Windows still supports
applications in an MS-DOS task.  And of course when it comes to "PC"
hardware, compatibility has produced some incredible kludges we're
still saddled with, the pitiful instruction set architecture being
perhaps the worst, with lack of standard video interface being a
close second.  Apple's OS/X systems are thus no worse in regard to
evolutionary features than Bill Gates' Windows systems.

Unix got several things right, most notably treating (almost) all
data sources/sinks as "files" that are accessed via the same mechanism.
Plan 9 explores the consequences of pushing that idea even farther,
while dropping some of Unix's less desirable features, but is not
bolted onto anything except existing hardware.

Re: [9fans] missing compilers?

[jkw]

In article <3D4F8DF3.18613.27913A3F@localhost>, Andrew Simmons wrote:
>> I'll second Russ's comment: why not upgrade?

> bolted on top of a "new" operating system which is based on
> Carnegie-Mellon's 1980's version of an operating system from the
> 70's.  At which point you start to think that even Bill Gates never legged
> you over to quite this extent, and start looking at Windows machines.

Ah.  I see, you're from the VMS camp, then?

Re: [9fans] missing compilers?

[Andrew Simmons]

> I'll second Russ's comment: why not upgrade?


Partly a matter of cost, and partly a matter of taste. OS X runs like
a pig
on last year's top of the line model, and is none too flash on the
current top
of the line. So you'll need new hardware, plus you just know that
none of
your old applications will work, so you'll have to buy new versions.
And
get new drivers for your peripherals, if you can. And having shelled
out thousands of dollars, what do you get? A confusing new user
interface
bolted on top of a "new" operating system which is based on
Carnegie-Mellon's 1980's version of an operating system from the
70's.
At which point you start to think that even Bill Gates never legged
you
over to quite this extent, and start looking at Windows machines.

Re: [9fans] missing compilers?

[Russ Cox]

geoff explained my reasoning.
if someone wants to go through the torture to get drawterm
running on old macs, please do.  otherwise, it works on os x
and you are welcome to switch in order to run it.

i'm not going to address your trolling.  take it to comp.os.macosx
or better yet comp.os.religion.windows.

russ

Re: [9fans] missing compilers?

[Sam]

Ahh,

<sincere>
It's so nice to have you back, Russ.
</s>

;)

Sam

On Mon, 5 Aug 2002, Russ Cox wrote:

> geoff explained my reasoning.
> if someone wants to go through the torture to get drawterm
> running on old macs, please do.  otherwise, it works on os x
> and you are welcome to switch in order to run it.
>
> i'm not going to address your trolling.  take it to comp.os.macosx
> or better yet comp.os.religion.windows.
>
> russ
>

Re: [9fans] missing compilers?

[Douglas A. Gwyn]

Geoff Collyer wrote:
> Did 68k-based Macs even *have* Ethernet interfaces?

Why not?  Even the Apple II had an Ethernet interface (option).
It's true that Apple had a "NIH" mentality and tried to keep the
Macintosh system relatively closed for a long time; remember
AppleTalk protocol (used where we would these days use TCP/IP)?

> I'll second Russ's comment: why not upgrade?

Maybe he's looking for some way to use his old equipment,
just as many people apparently use antique PCs with Plan 9.

Re: [9fans] missing compilers?

[Geoff Collyer]

> As long as we're talking about Plan 9 on Apple hardware, what about
> the Motorola 68x based macs?

It's reasonable to consider porting Plan 9 to machines that can run OS
X only because the hardware is documented at least in the source code
for Darwin (and other systems, such as OpenBSD), which is pretty
likely to be correct and current, as documentation.  I'm told that in
the past Apple operated on the same principle as NeXt: that details of
the hardware such as ROM contents could be changed on a whim since the
company could then just tweak the OS to match.  Thus the hardware was
effectively undocumented, or at least underdocumented, from the point
of view of someone writing code for the bare hardware.  One could use
the OpenBSD mac68k port as hardware documentation, but I think that
porting to the 68k-based Macs would be, at minimum, painful and
probably the resulting port would run on some machines of a particular
model and not on others of the same model (that was true of the Plan 9
port to the NeXt).  And aren't 68k-based Macs awfully old?  As I
recall, Apple started shipping Power PC Macs in 1994.  Did 68k-based
Macs even *have* Ethernet interfaces?

> Also, I see you mentioned Inferno under OS X.  For those of us who haven't
> upgraded yet, would it be possible to port it to the classic Mac os's?

I'll second Russ's comment: why not upgrade?  If it's because your
hardware is old enough that OS X won't run on it, I think you're just
stuck.

The internal differences between OS X and all earlier Mac OSes are
enormous.  OS X is essentially a recent BSD Unix system, with all that
that implies: multiprogramming with pre-emptive scheduling, memory
protection of the kernel from user processes and user processes from
each other, POSIX threads, ANSI/POSIXy C library.  It's my
understanding that earlier Mac OSes had none of these.  (Disclaimer: I
avoided earlier Mac OSes, and did hear the rumour 10 years ago that
System 7 did some minimal memory protection, but heard later that it
wasn't effective.)  Even Windows has some of these, to varying degrees.
I'd expect porting drawterm to earlier Mac OSes to be painful, if it's
even possible.  OS X knows how to cope with 3-button mice; did earlier
Mac OSes?  If you're stuck with the 1-button Morse-code-key mouse,
Inferno and drawterm would be awkward to use.  Plan 9 would be too if
older Mac hardware makes it impossible to replace the Morse-code-key
mouse with a 3-button one.

Re: [9fans] missing compilers?

[Andrew Simmons]

> or anyone who cares could upgrade to os x.

I wouldn't use the word "upgrade" myself.

Re: [9fans] missing compilers?

[Jack Johnson]

Russ Cox wrote:
> or anyone who cares could upgrade to os x.  i don't see why we
> should bother catering to people running old systems.

We're running out of landfills in China for our old Macs.

-Jack

Re: [9fans] missing compilers?

[okamoto]

[-- Attachment #1: Type: text/plain, Size: 230 bytes --]

Probably, people still has an image that older Plan 9 could run
on a poor machine very fine.   If we want to use present, release 3
included, Plan9 effectively, we need AGP at least, and more 
than 64MB memory.   ^_^

Kenji

[-- Attachment #2: Type: message/rfc822, Size: 1845 bytes --]

From: "Russ Cox" <rsc@plan9.bell-labs.com>
To: 9fans@cse.psu.edu
Subject: Re: [9fans] missing compilers?
Date: Wed, 31 Jul 2002 13:10:32 -0400
Message-ID: <97523145b64b70fe18f0ef4980864d66@plan9.bell-labs.com>

> So lack of processes should be possible to over come fairly easily.
> ... perhaps a deal could even be cut with the Tenon people.

or anyone who cares could upgrade to os x.  i don't see why we
should bother catering to people running old systems.

Re: [9fans] missing compilers?

[Russ Cox]

> So lack of processes should be possible to over come fairly easily.
> ... perhaps a deal could even be cut with the Tenon people.

or anyone who cares could upgrade to os x.  i don't see why we
should bother catering to people running old systems.

Re: [9fans] missing compilers?

[Todd Olson]

In article <fdd5414e6fdd87fd1d8ae66a30f2048e@vitanuova.com>,
 rog@vitanuova.com wrote:
> 
> it would be very difficult as classic macos doesn't have proper processes,
> which inferno relies on.
> 

Over 10 years ago these people
http://www.tenon.com/
create a Mac program that behaved as UNIX
It even ran on a MacPlus.
It eventually involved in to a nice system
that you could do many things both inside the UNIX environment
and outside.

They stopped marketing it when MacOSX came out.

So lack of processes should be possible to over come fairly easily.
... perhaps a deal could even be cut with the Tenon people.

Regards,
Todd Olson
Cornell University

Re: [9fans] missing compilers?

[rog]

> Also, I see you mentioned Inferno under OS X.  For those of us who
> haven't upgraded yet, would it be possible to port it to the classic
> Mac os's?  I know that it'd be difficult because of all the resource
> stuff, plus there's no command line-like place for output.  (There's
> always MPW's Worksheet, but I'm guessing that not many people use it
> anymore.)  Anyway, has anyone even looked at this?

it would be very difficult as classic macos doesn't have proper processes,
which inferno relies on.

Re: [9fans] missing compilers?

[Ben]

As long as we're talking about Plan 9 on Apple hardware, what about
the Motorola 68x based macs?  Shouldn't one be able to port it to one
of these machines, or does Plan 9 require the PowerPC stuff?

Also, I see you mentioned Inferno under OS X.  For those of us who
haven't upgraded yet, would it be possible to port it to the classic
Mac os's?  I know that it'd be difficult because of all the resource
stuff, plus there's no command line-like place for output.  (There's
always MPW's Worksheet, but I'm guessing that not many people use it
anymore.)  Anyway, has anyone even looked at this?

Re: [9fans] missing compilers?

[Geoff Collyer]

The powerpc compiler is there but no binary is supplied, so you need
to compile qc, qa and ql yourself.

I have inferno running on OS X.  It's still got some quirks and seems
to work better on 10.2 than 10.1.

A Plan 9 port needs at least a driver for the relevant Ethernet
controller (Apple's GMAC is built in to the Power Mac, as I recall).
For pre-VGA console access, it appears that one needs to use Open
FirmWare's console get and put routines since recent Macs don't seem
to have an accessible serial port.  Keyboard and mouse are normally
USB, which sounds like a pain.  It looks like there is a z8530 DUART
connected to a modem chipset, but the only external connector is a
phone jack.  I'm not sure which interrupt controller and MMU are used,
but all this information should be available in the Darwin or OpenBSD
sources.  The supplied VGA is usually ATI or Nvidia.  Disks and
CD-ROMS are normally IDE.

On the other hand, the Power Macs at least have several PCI slots, so
one could perhaps get an initial port running quickly by inserting
cards that already work on the PC, for example Intel i82557 Ethernet
or Mylex SCSI.  One can substitute Wavelan PCMCIA cards for Airport
cards.

The Open FirmWare boot loader can load AIX XCOFF binaries, which it
appears ql can generate.  As far as I know, qc doesn't exploit Altivec
instructions, so you probably won't see much advantage with a G4 as
compared to a G3.  3rd edition contains an mpc port and 4th edition
contains an mtx port.  One of them is probably a reasonable base to
start from, depending mainly on MMU and interrupt controller.

Re: [9fans] missing compilers?

[Scott Schwartz]

Speaking of compilers, what's the status of the PowerPC?  (In particular,
the G4 Macintosh.)

Re: [9fans] missing compilers?

[forsyth]

>>somewhere something about a compiler for the Hobbit processors...

the hobbits were hobbled years ago.

[9fans] missing compilers?

[Ben]

I read in the 2c manual page that there should be a compiler for the
AT&T 3210, but cannot find it anywhere, in source or binary form. 
Does it still exist?  Also, (I'm probably mistaken...) I read
somewhere something about a compiler for the Hobbit processors...
Anyway, should the 3210 part be removed from the manual page, or
should the compiler still be available?