A Technical History of Apple's Operating Systems

An anonymous reader writes "As part of his 1680-page book Mac OS X Internals: A Systems Approach, Amit Singh of kernelthread.com wrote a very detailed technical history of Apple's operating systems. Since he had to cut down on the history chapter because of the book's already too-large size, most of this chapter didn't make it to the printed book. Singh has made available the history chapter as a free PDF. The file is 140 pages long, and is generously filled with figures and screenshots. It starts with the internals of the original Apple I and goes through a tour of all operating systems Apple dabbled with, including internals of A/UX, Lisa OS, and such. It even covers details of outside influences like the Xerox Alto, STAR System, Smalltalk, and Sketchpad, and closer to home things like Mach, NeXTStep, and OpenStep."

Apple ][

One of the coolest things about the Apple I and Apple ][ was that Apple Computer included the schematics for *all* of the motherboard and CPU design. Everything was documented so that users could build interfaces with both the software and the hardware with a minimum of fuss. So, even though Amit Singh calls the manual included with the Apple ][ as a "preliminary manual, it was remarkably complete.

Despite how far we've come, there are time I really miss my old Apple ][.

Re:Apple ][

I think that's what they wanted you to do. Even if you didn't buy a Sony TV, a substantial percentage of people will, so Sony wins by making you buy a new TV instead of fixing your old one. Or they win by charging you an outrageous price for the schematics and repair manuals.

And the reason that they do this is that they (and you) don't have to pay the real cost of disposing of the old TV. Instead of recycling the TV and reclaiming all the materials, you'll probably just toss the old TV in the trash. And the hazardous chemicals will leak into the soil. Our descendents will have to clean that shit up eventually, which will cost tons of money. But we don't have to pay that, so we just go get a new TV cheap.</rant>

Re:Apple ][

That doesn't change the fact that assuming that the future will magically fix all the problems is entirely irresponsible.

In Soviet Russia...

In Soviet Russia YOU fix old TV's.

Uh, wait.

Re:Apple ][

was that Apple Computer included the schematics for *all* of the motherboard and CPU design.

God, we have come a long way haven't we - now Apple will cease & desist you for linking to their Service Manual.

God, how I miss the old Apple :-(

Re:Apple ][

If companies would package their products to include tech specs and schematics, people who don't want to mess with their purchased property wouldn't have to, but the people who want to modify, repair, or extend their purchased property could do so with ease.

And don't give me the old, tired, whiny excuse that people would simply build their own from the specs they got from a friend. It's not true. As you alluded to, most people aren't hobbyists and don't want to be bothered to build their own. And there isn't a problem from a commercial competitor, either, since patents and copyrights are there to protect against this exact form of abuse. There are adequate legal protections against ripoffs.

Companies should be required to include specs with every electronic and mechanical device they sell, whether it's as small as a wristwatch, or as large as a car.

Re:Apple ][

My Aged Mum, now in her eighties, bought the first Apple ][ ever sold in her small Southern city and shortly thereafter traded up to the ][c. She was an artist by trade. The first thing she did was to construct a couple of cables that she needed for her work (video was one that I recall). Then she sat down with the manuals, learned Applesoft BASIC, and wrote a program or two that enabled her to generate patterns for complex weaving with a large loom. Eventually she acquired an interface that allowed the Apple to actually drive the loom--it was a complicated system of switches and relays that raised and lowered the various harnesses or frames on the loom. She did all of this when she was past fifty and with no prior training at all, though she was regular in attendance at users' group meetings once a users' group was formed.

I still have (and treasure) bits of cloth of complex, intricate design, created and produced with the aid of that Apple. She truly made it an extension of herself.

Not as good as the Beeb though

For me, the best 8-bit computer ever was the BBC micro - I doubt it ever gained any traction over here in the US, but *man* was that a well-designed and elegant machine.

The OS was fully vectored and modular, the BASIC language had procedures and functions, as well as a built-in assembler that could access BASIC variables, but the hardware design was what made it stand out. It had every i/o port under the sun - serial, parallel, "user i/o", other dedicated ones for a network (Econet), to support floppy disks and hard disks, and even plug in a second co-processor (there were 8086, Z80 and 32000 variants I think). You could get Pascal and C for it, and it supported 80-column text on a monitor.

And to bring it slightly back on-topic, the documentation was simply excellent - the "Advanced user guide" told you just about everything you needed to know about the machine, from the event i/o to interrupt-programming, documenting the OSxxx calls, and all the port i/o etc.

Nothing since has come close to the flexibility of that machine given the design limitations at the time, and it's a tribute to the designers.

Of course, such largesse can be abused [grin] See My first and only virus-writing incident [slashdot.org] ...

Simon

Those were the days...

I remember making "awesome" games in the 40x40 graphics mode. Not too easy to make a game in a couple hours anymore ;)

http://religiousfreaks.com/ [religiousfreaks.com]

What I want to know is

Who liked MacOS prior to OSX?

What were you smoking?

Where can I buy some?

Re:What I want to know is

Actually, it wasn't that bad... in the earliest iterations, it was miles above what Windows could provide, and for ordinary users, it was rather slick. I remember as a 16-year-old the day I saw the Macintosh 'Classic' for the first time when my mother brought it home from work for the weekend (the Mac Classic form factor was fairly brand new then) - compared to the Windows 2.0/DOS rig on my parents' shiny new Amstrad "2286" (remember DOSSHELL? - Windows 2.0 it was pretty much like that)? The Mac blew away Windows/DOS in usability, presentation and performance. It was damned slick.

Of course, time went on, and things changed radically since then, but Mac UI development was, in its early days, miles beyond what Microsoft could muster.

Now - why MacOS decided to stick with the same setup in spite of Win95/98? Dunno.

/P

Re:What I want to know is

It's not so much that Apple "decided" to stick with the classic Mac OS for so long despite Win95/98 as much as that they were just in a total mess in terms of their next gen OS stuff. Apple spent a lot of the 90's trying to figure out a new Mac OS, and a lot of the future was supposed to be in the original projects codenamed Copland and Gershwin (the original Mac OS 8 and Mac OS 9...... what eventually came out as Mac OS 8 and Mac OS 9 had little to do with this). Copland was originally supposed to be a totally modern OS, and Gershwin would apparently have had even more radically new elements, a lot borrowed from the Taligent collaboration with IBM on an OS codenamed "Pink". But none of this ever panned out, and all Apple could do was release the commercial Mac OS 8 and Mac OS 9 OS's as stopgaps. Apple had considered buying BeOS once it became clear that the internal Copland project wasn't working out, and they ultimately ended up buying NeXT. For all intents and purposes, Apple became NeXT, and Mac OS X can be seen in many ways as the ultimate development of NextStep, rather than the classic Mac OS.

Re:What I want to know is

Actually, looking through the Mac OS X Internals PDF, there's a whole section there about Apple's quest for a new OS........ this goes into a lot of detail about all the options Apple considered. This is actually some great reading if you're curious about all of this.

I did

Daring Fireball wrote about this recently [daringfireball.net]. Here's the most important quote of the article:

The difference between the old Mac OS and Mac OS X isn't that it used to suck but now it's great. The difference is that Mac OS X's appeal is broader; it is good in more ways than the old Mac OS was.

Yeah, I did use and like Mac OS 9, Mac OS 8 and System 7. I did smoke lots of weed, but that had nothing to do with it. There are two things to consider: First, it went up against crap like Windows 3.11 and Windows 95. Second, it was the prettiest, most easy-to-use OS, even with cooperative multitasking and lack of memory protection.

Mac OS X added a lot to what makes a Mac great, but Mac OS 9 had a lot going for it, too.

Re:I did

... it was the prettiest, most easy-to-use OS, even with cooperative multitasking and lack of memory protection.

Memory protection used to be explained in the following way:

• UNIX - if a program needs more memory, the system gives it more memory
• MacOS - if a program needs more memory, the system tells you and you have to give it more manually and try again.
• Windows/DOS - if a program needs more memory, your computer just crashed.

For all practical purposes this was the state of things for many years.

Re:I did

For all practical purposes this was the state of things for many years.

It's a rather misleading description, though. More accurately:

• UNIX/Windows NT/OS X - if a program needs more memory, the system gives it more memory; if there is no more memory to give, the program is terminated.
• Windows 9x - if a program needs more memory, the system gives it more memory; if there is no more memory to give, your computer crashes.
• MacOS 8/9 - if a program needs more memory, the system tells you and you have to fiddle around with a fussy little dialog box to give it more manually and try again, at which point another program will complain that it no longer has enough memory. Repeat ad infinitum, all the while gritting your teeth and reciting the mantra "this is better than Windows, this is better than Windows" until you almost believe it.

Re:I did

It's a rather misleading description, though. More accurately: UNIX/Windows NT/OS X

Umm, the memory management issues changed long before OS X existed and this predates even Windows NT for the most part. I'm not sure exactly what you're trying to describe, but you fail to describe either the state of the art now, or the situation as it existed in the past, but rather have presented a muddled, mix of both, while leaving out most of the concepts of modern memory management. "if there is no more memory to give, the program is terminated" is certainly not the case with any modern UNIX or with OS X, as it jumps to swap and then frees memory from other systems according to how they are "niced" among other things.

Repeat ad infinitum, all the while gritting your teeth and reciting the mantra "this is better than Windows, this is better than Windows" until you almost believe it.

The first computer I ever personally owned was dual motherboard, dual processor 66mhz ppc and 486/66 simultaneously running both Windows 3.11 and MacOS 7.x (with a cool key combo to swap the input and display and some nifty utilities to copy and paste between them). I'm about as close to an impartial observer at the time as you could have ever had. The fact was, Windows memory allocation was in theory, much better than MacOS, but in practice was so unstable that it caused an even bigger problem than it solved. If you don't remember this than you either never ran both side by side or you are looking at the past with rose tinted glasses.

Re:What I want to know is

I loved Mac OS 8. I never really used OS 9 but by then it was going up againist win2k.

Mac OS 8 went head to head with win98. The only better choice at that time was BeOS . Yet another good OS killed by an illegal monopoly of a bad OS.

Hell BeFS featured a database File System of the likes MSFT still can't create. and they did it on hardware that even Linux would require recompile and lightweight window manager to run on.

MSFT to this day is still trying to copy cool features found in competitor's now old products.

Re:What I want to know is

The features that MacOS Classic lacked were mainly:

1. Good multi-user support.
2. Pre-emptive multitasking.
3. Protected memory

If you wanted these three, you had two choices; UNIX or Windows NT[1]. Other consumer operating systems didn't have them. Windows gained Pre-emptive multitasking with Windows 95, but it didn't get the other two until MS abandoned 9x in favour of NT. BeOS didn't have the first, but did have the second and was quite popular with Mac users.

What it did have was a heavily Raskin-influenced GUI, which left pretty much anything else in the dust when it came to usability. NeXTStep was in the same area, but much more expensive.

[1] Or VMS and a few others if you had a huge budget.

Amit's Book

I did a technical review of the book, and I can thoroughly recommend it (I got a free copy). It's very technical in places (lots of code snippets) but does a very good job of explaining the 'why' as well as the 'what' and 'how' of XNU.

First, the Earth cooled.

Then, the UNIX came, bit it got too big and fragmented, but it didn't die out, and turned into BSD.
Then Steve Jobs came, and he brought forth NeXTStep.
And then Apple bought up NeXTStep, added some more BSD, and gave it some pretty clothes and called it OS X. I couldn't believe it. They opened the closet, took out the best eye candy, and walked straight into town...

Archeological dig

From TFPDF:

Lisa was discontinued in 1985. In September 1989, Apple buried about 2700 Lisa computers in the Logan landfill in Utah. The value of the computers had depreciated so much that the tax break received from scrapping the computers resulted in more money for Apple than could be obtained by selling them.

Anybody feel like digging? :)

Re:Archeological dig

I'll go with you, just so long as we make a video of it like these guys did when they dug up E.T. the Extra-Terrestrial [keithschofield.com] for the 2600 [wikipedia.org]. ;)

quick obselescence

I took OS comparison course in MIT's business school some years back, and must say of all the courses I took in computer science, that one became the least useful the quickest. OS's seem to be changing faster than computer languages, and much faster than good SE techniques like design patterns.
I'm guessing that it is because the underlying hardware is changing rapidly- many hardware sectors increase in size and performance a magnitude every five years, making some resource algorithms less meaningful. Plus novelties like flash, MRAM, cores, cells, and GPUs, etc. all have to be integrated in.

1680-page book ?

This is what happens when you get a contract that says you're paid by the word.

Lisa OS

My CS Prof. at the time (Summer of 1982 or 1983, an old retired IBM'er who worked on the first computers for the Military) had a daughter that worked for Apple on the Lisa project. He had a pre-production model on his desk with a serial number under 300. She needed Steve Jobs personal okay to send him the computer for his testing. (So I was told) I remember it was the coolest thing I'd every seen back then. We took the cover off and his daughter's name was one of the names inscripted on the inside cover. Blew away the Apple II & Trash-80's we were using at the time.

No wonder the book is 1680 pages

This article is information overload in the extreme. What does Apple II DOS have to do with OS X? Or why Wozniak chose the 6502 over the 6800? Or the Apple III SOS or Apple II Prodos? Or Apple transitioning to PowerPC chips in 1994? Some of the newer stuff is interesting to know, for historical reasons, like the failed OS development projects that led up to OS X, but there's no way this should have been 140+ pages. It doesn't bode well for the rest of the book.

Why it didn't appeal to christians...

"The Apple I was introduced at a price of $666.66." Coincidence? Conspiracy? Nah, just a good price.

wished for more about A/UX

As the new maintainer of the A/UX FAQ [dyndns.org], I keep hoping to learn more about it. Unfortunately the author didn't bring up anything I didn't already know. That said, the page or two he had is a good summary for those that have never used A/UX before.

I bought this book - it is very good.

I pre-ordered this book and I received it maybe 10 days ago. It is a very good book.

Excellent form of promotion

I think it would be cool if more authored did this - releasing the "deleted scenes", so to speak, of their works for free, as a promotion for what they kept... especially if they were still high quality. It probably increases sales quite a bit (I wasn't even considering buying this book until after looking at the sample), and gives something useful directly to the community.

What was the book written in?

Word? The random choice of fonts, and crappy layout makes it clear that this highly skilled engineer, "Doesn't know LaTeX!"

Being too greedy?

I was always a fan of BeOS. A presentation and demonstration of it was part of the interview process that got me my first consultant position. Always thought was a shame it didn't make it. Now, if that chapter is to be believed (p100 of the pdf), they could have were it not for the boss being too greedy. After managing to negotiate upwards from an initial valuation by Apple of $50m up to an offer of $200m, he still tried to get more and got no deal at all.

That's a little heart breaking, actually. I've been lucky enough to avoid the whole "my company's gone bust" thing, but what I've seen others go through isn't nice. It would really piss me off to find out that the boss, with $200m on the table and only $20m having been put into the company, still was too greedy / crap at negotiation to take it.

"Ten times return on our investment? Ha! I think we'll do better trying to compete with Microsoft in the OS market!"

Yeah, real smart...

Anyone know whether Jean-Louis Gassee really could have accepted the $200m offer and closed the deal, e.g. if even Apple didn't end up putting BeOS at the heart of their OS strategy?

Re:Being too greedy?

A bunch of us engineers spent the time between Copland's abrupt halt and the NeXt acquisition trying to figure out which operating system the company was going to try. There was a lot of experimentation with MkLinux and some talk about beefing up A/UX but the biggest buzz was coming from the BeOS. A few of us made the pilgrimage to Menlo Park, saw their presentation, and were mightily impressed with its performance, but we agreed that the lack of available consumer applications made it a non-starter.

From the time Copland died in the summer of 1996 until we got laid off in March of 1997, we waited for the Big Decision and learned a lot about UNIX-based operating systems because we knew that's where the company had to go. NeXt and Steve Jobs's return were complete surprises. Smartest move Gil Amelio made--just as was Steve's immediately getting Gil out of the way and resuming leadership. Apple's customers needed a reason to believe and Gil only provided silence. As one Rumor-Monger wag said, "he couldn't market pussy in a prison."

Is this history in the right book?

Look at it this way:

Suppose you bought a book about Ruby and it had a 140 page introductory chapter that started with the history of FORTRAN and Algol and worked all the way up through Perl and Python and Ruby. Suppose it went off on tangents about things that Niklaus Wirth and John Backus did. Suppose it talked about various programs of historical importance (like the first compiler and PLANNER and so on). Suppose it talked about programming paradigms which don't relate to Ruby, like the development of Prolog.

Now all of this is interesting, but it's out of place in a book about Ruby.

Surprised at so little mentioned about GSOS

GSOS and the Apple IIGs was quite the sophisticated platform and I'm surprised the author left out the little bit about how Apple alienated a large majority of its customers thanks to the Apple IIGS. The GS was my first "real" computer as a kid. My parents had and I had dabbled with an Amiga long before the GS, but the GS was my first real "work" computer where I did word processing and more with it. It was also my entry point into the early days of the internet and the first computer I ever upgraded with double density disk drives, a 40mb hard drive, various dial up modems, etc.

For me the AppleIIGS was really the "begining" of my current career in the computer industry. It was also a really slick operating system. But the most significant impact the AppleIIGS had on the market was it was the start of Apple's trend of abandoning old technologies. Almost as soon as the AppleIIGs was released, Apple had abandoned it and the Apple II platform for its new Macintosh systems. When Apple did this they abandoned the large majority of their customers. The early Macs were relatively expensive versus the bargin prices on Apple IIs, and a number of schools were deeply invested in the Apple II platform.

When Apple abandoned the II with the GS it was the start of the first major shift in the personal computer marketplace. A number of Apple customers felt gilted by Apple so they began to look for alternatives. Compared to the expensive Macintosh, the relatively cheap PC clone industry seemed like a huge bargin. It was at this moment that Microsoft really took control of the Operating System/platform market as a large portion of Apple's customer base abandoned the company and switched over to PC clones powered by Microsoft's Operating Systems. In truth, it has only been with Mac OS X and their Mactel platforms that Apple has truly succeeded in significantly expanding their marketshare since the AppleIIGS fiasco.

As I said, for an operating system and product that had such a profound impact on the future of Apple, I'm surprised to see so little mention of the AppleIIGS and GSOS.

Re:who cares?

MAC OS is cool and all, but is this really necessary?

Too early for existential type questions. I think I need my coffee :)

Re:What a load of crap

In your post, it says "I give up sometimes I really do..."

NO YOU DON'T, YOU JUST WASTED YOUR TIME explaining, "This page is intentionally left blank" SO YOU HAVEN'T GIVEN UP!!! :P

Re:What a load of crap

I don't understand what you're complaining about.

Re:What a load of crap

What it should really say can be generated by the Python

def blankpage():
print "This page intentionally left blank except for a message stating `",
blankpage()
print "'"

(damn Slash indentation destroyer) but I think they renormalised it.

Re:1680 page book?

Those are all works of fiction though. I'm talking about a real concrete book about science.

[hehehe yeah I am joking].

Chances are if your book is 1680 pages you should divide and conquer that sucker.

I mean I could write a book called "All there is to know about computers" and cover software, hardware, design, engineering, algorithms, etc, in one huge 32,618 page book. That doesn't make it a good idea.

Tom

Re:What a load of crap

Just a piece of advice for you: don't ever try to work for the government, okay?

Re:Singh is an idiot - so many errors

Why is the parent poster getting modded up as informative? I'd trust Amit over what appears to be an obvious troll. I'd also trust the 6500 spec sheet and the original Apple manual that I managed to dig up.

For example, it says in the Spec sheet "Addressable memory range of up to 65K bytes", "On-the-chip clock options: Crystal time base input", etc:

6500 data sheet [6502.org]

"Microprocessor Clock Frequency: 1.023 MHz"

Apple I Manual [computerhistory.org]

etc.

Re:Singh is an idiot - so many errors

Singh may be really good at the newer stuff, but the old stuff he obviously did NOT dig into the hardware or software first hand. There ARE LOTS of of errors, half-truths, and very important omissions in this chapter! Let's list a few:

• The 6502 had an alleged crystal-control capability, but in every 6502 board I've seen (almost a dozen), every one of them uses a separate clock oscillator.
• You can only get 16 addressing "modes" if you count each register as a different mode. And the indirect Y mode I don't think anybody ever found a use for. More like 6 usable modes.
• Dunno where he got the 65 cycles reserved for refresh stuff. The CPU had no refresh mode built in.
• The Apple ][ came with a DC power cord?
• The Lisa could not have a on-chip MMU, as there was no MMU available for the original 68000.
• The discussion of memory management on the Lisa is all wrong. The Lisa used a 68000 CPU, which had a major flaw-- memory faults were not restartable, plus there was no virtual memory management unit on the CPU or available. SO virtual memory was extremely constrained:: Apple had to design their own (slow and very limited) MMU, AND page faults were completely disallowed! A program had to obey a certain page-touching protocol to succesfully run. All stack expansion had to be first pre-flighted by touching a lower stack location, otherwise the system would hang. Not a pretty state of affairs.
• The writer obviously has never tried using or programming the Lisa Office System. Some tidbits:
  • It took about 30 seconds for the calculator to pop up. Then when you typed "1+2", it took about another 30 seconds of disk-whirring for the system to load the SANE numeric library and cough up "3".
  • Any system API call took nearly forever. You dared not poll for input very often as that would chew up a lot of time and slow down the whole system. But you couldnt try polling every so often, because just the call to get the system tick count took forever. Catch-22 ala big-time.
• The UCSD p-system was not command-line driven, it used top-line menus and single keystrokes.
• The UCSD editor was "modal", as it did have insert nad delete moes, but perhaps more importantly it was perhaps the first FULL SCREEN EDITOR accessible to the masses. Speedy too, as long as your file would fit in RAM.
• The Mac did not have a four-voice audio synthesizer. there was only one 8-bit audio D/A. Any multiple voicing had to be faked in software.
• Very cursory and obviously second-hand discussion of the Mac Toolbox API's. Not a mention of the revolutionary concepts introduced, such as the memory manager, memory ptrs, memory handles, resource forks, system hooks, drivers, etc...
• Multifinder was an outgrowth of Andy Hertzfield's "Switcher", and very very slick for a cooperative multitasking non-memory protectred system enhancement.
• Opinions may differ, but to many many people, System 7 was in no way a gigantic leap forward. Some of its "features":
  • Roughly a doubling of the basic API's, all of them implemented as RAM-hungry patches.
  • A Color QuickDraw, which was SEVENTEEN times slower at drawing text than the previous QuickDraw in ROM. Andy Hertzfeld eventually released a recoding of this to speed it up considerably.
  • A completely redone and completely unusable sound management API.
  • A basically unsable virtual memory system, trying to squeeze memory through a slow 8-bit SCSI port.
  • Proprietary "publish" and "subscribe" info sharing API's, which nobody ever used.
  • A rewritten Finder, with many many faked features, such as font and system resources presented as fake folders of "files".
  • Poor support for TCP/IP, threading. No support for SLIP or PPP.
  • A completely redone networking stack, based on all new, slow, and lousy code.
  • Much higher memory usage, very slow in places.
  • Many of us stayed with system 6 as long as possible, even Apple re