If Apple never did the Mac, would the Apple IIGS have thrived?

[liam_on_linux]

The 65C816 was a dead end, I'm afraid. It was a fairly poor 16-bit chip, and the notional successor, the 65C832, was never made. It only existed as a datasheet:

https://downloads.reactivemicro.com/Electronics/CPU/WDC%2065C832%20Datasheet.pdf

Backwards compatibility is really limiting. Look how long it took the PC to catch up with mid-1980s graphical computers, such as the Mac, Amiga or ST. Any of them was frankly far ahead of Windows 3.x, and it wasn't 'til Windows 95 that it could compare.

Innovation is hard. Everyone tends to overlook the Lisa, which is the machine that pioneered most of the significant concepts of the Mac: not just the GUI, but a rigorously and completely specified set of UI guidelines, plus a polished, 2nd-generation GUI.

(Xerox's original was very Spartan. No menu bars, no standardised window controls, no standardised dialog boxes, etc. It was a toolkit for writing GUI apps, and a fancy language to implement them in.

Apple added a _lot_. But the first version was, just like the Xerox Star, way too complicated (hard disk! Multitasking!) and *way* too expensive.

It took a second system to get it right, and it took cutting it back *HARD* to make it affordable enough so people would notice. Yes, sure, 128 kB wasn't really enough. One single-sided floppy wasn't enough. But even so it was $2500. It had to be pared to the *bone* to get it down to a quarter of the price of the Lisa.

It was a trailblazer. It showed that a single-user standalone GUI machine was doable, and worth having, and could be just about affordable.

Just 9 months later, a 512 kB model was doable for only $200 more. Tech advanced fast back then.

They simply could not have done a IIGS at that kind of price point in 1984. It wasn't possible. The Mac was only barely possible. The other new 680x0 personal computer of 1984, the Sinclair QL, had 128 kB too.

If there'd been no Mac, the GS wouldn't have had its GUI. The GUI was a re-implementation of the Mac one. Without that, it would have just been a slow kinda-sorta 16-bit machine, released a year and 2 months after the Amiga 1000 – which was $1300 but which had much better graphics, comparable sound, a full multitasking GUI, and a 7.1 MHz 68000 – a much more capable chip.

Or the Atari ST, which was another full 68000 machine, with half a meg of RAM, and a GUI, and was (unlike the Amiga) usable with a single floppy because the OS was in ROM... and which was $800 in June 1985.

There is more to the universe than just Apple.

In the gap between the Lisa and the Apple IIGS, IBM released the PC-AT, which my friend Guy Kewney, perhaps the most famous IT journalist in the UK then, called "his first experience of Raw Computer Power". His caps.

The year after that, Intel released the 80386, a true 32-bit chip. The same month as the IIGS, Compaq released the Deskpro 386, the first true 32-bit PC. Sure, $6,500 -- but vastly more powerful and capable than a 65C816.

The IIGS was a gorgeous machine. I was at the UK launch. I wanted one very badly. But bear in mind that the Apple II was _not_ a successful machine in Europe -- it was was too expensive. A $1000 computer in 1977 was no use to us: that was more than the price of a car. We got Sinclair ZX80s and ZX81s, the first £100 computers. :-)

So outside a few countries, the IIGS had no existing catalogue of software and so on. Neither did the Amiga or ST at launch, but they'd been around for over a year by the time the IIGS appeared, and they had amazing best-of-breed apps and games by then.

Comments

[history_monk]

Yup, the 65816 had a serious problem, which the 8086, amazingly enough, lacked. The extension of memory addressing to 24-bit (16MB) was done with an 8-bit register ("DBR"), which formed the upper 8 bits of the memory address. That meant that memory was addressed in 64KB segments, which had fixed, non-overlapping addresses. Copying between segments required changing the DBR twice for each 16 bits you copied AFAICS. This was slow!

The 8086/88 also divided memory into 64KB segments, but they could start on any address that was a multiple of 16, so copying was quite straightforward, given the spare segment register provided for this purpose. This was far more practical than the 65816, or the Zilog Z8000, which had an identical problem. A flat address space, as offered by the 68000, was far better, of course.

[liam_on_linux]

Thanks for the validation! :-)

Oddly enough the single best source of info I have come across for the limitations of the 65C816 was a single passing comment on Hackernews.

Might have been one of these...

https://news.ycombinator.com/item?id=22237661

https://news.ycombinator.com/item?id=27209469

[history_monk]

Aha, the second of those points out something I didn't know, which is the ability to store a 24-bit address in zero page and indirect through it. That relieves the memory addressing problems considerably, although it's going to be fairly slow.

I was working on the BotStik software for the Apple II and BBC Micro in 1984-86. When the Apple //c came along, we concluded it was a marvel of misplaced ingenuity, and continued with our MS-DOS product, which lasted into the 1990s.

[tpear]

Certainly the '816 was not a huge success in the home/desktop computer market — but given it looks like you can still buy them off the shelf today, I suppose it must’ve found a niche somewhere?

The IIGS deliberately limited the speed of its ‘816 to avoid it showing up the Mac; I believe it also had a colour GUI way before the Mac.

I wonder what might’ve happened if Lisa had more success. I think it’s quite a different beast to the original Mac and had some concepts UI that I, for one, would like to see even on today's GUIs… for example, the notion of creating a document by 'tearing off' the top sheet of an appropriate template.

[liam_on_linux]

but given it looks like you can still buy them off the shelf today, I suppose it must’ve found a niche somewhere?

I believe that there was a cut-down version for embedded stuff, and I guess that is why...?

The IIGS deliberately limited the speed of its ‘816 to avoid it showing up the Mac

I have often read this, but then I looked into it for a while, and found some good thorough analyses, and I am not sure it's 100% true. There may be a grain of truth to it, but the ][GS was I think the first mainstream device with one, and I think Apple was very concerned about yield and stability. Also, cost.

Secondly, it had to run slow in order to retain useful Apple ][ compatibility. Not just software but hardware too. You can put A][ expansion boards into a ][GS. Imagine if one of them cause the machine not to start?

I have a friend in Paris who has a ][GS waiting for me. I must go visit him again and get it. :-)

But the conclusion of others' investigations was: yes, there were faster models later, but at first, no, to get useful yield and good compatibility, they had to keep it slow.

I am no expert on Apple ][ stuff, though. Never had one. Barely touched them.

I believe it also had a colour GUI way before the Mac.

It's true, it did. Very low-res but it was there and it worked.

My impression of the GS OS is this:

If Classic MacOS was NT 3.x, then GS OS was Windows 3.

Sure, it looks almost the same, but it's horribly limited by the ancient clunky OS underneath. Yes, now, looking back, you could well think "wow, Windows 3 was amazing, and they upgraded it to 95 and 98! Why did they kill that? It was amazing, so much backwards compatibility!"

Well, yes, it was. But it was also deeply limited by the underpinnings and it could never have done all the stuff that the shinier OS did in time.

Classic itself became limiting and got pensioned off. Copland proved too much.

Enhancing the GS and GS OS would have been even harder AND it would have taken sales and R&D money off the Mac.

Apple was right to kill it, I'm afraid.

Selling it off would have been nice but that would mean selling off all the Apple ][ rights. Creating a competitor for itself. The Apple ][ card for the Mac LC came later. Apple still had interests in that stuff.

They did the right thing.

You know what I imagine as a better outcome? If the weird Atari/Commodore rivalry and stuff hadn't happened.

If Jack Tramiel stayed at Commodore, they never bought the Amiga and drove the development of the ST... CBM developed a 65C816-based C64 successor, and evolved it via something like the C65, with GEOS as its native OS. I reckon it had more legs and the C128 was a misstep.

And Amiga went its own way, maybe even bought by Atari and developed there. The Lorraine/Hi-Toro founders were ex-Atari and designed the amazing Atari 8-bit chipset.

Although in that imaginary world the ST might never have happened and that'd be a shame. The ST is what the QL should have been, with hindsight.

Navigating the alternate timelines is hard! :-D

I wonder what might’ve happened if Lisa had more success.

Well, yes, true.

I think it’s quite a different beast to the original Mac and had some concepts UI that I, for one, would like to see even on today's GUIs…

Also fair.

For me, I think, the fact that it had multitasking and that was removed, and then caused so much agony trying to add back again -- that's the key thing here.

for example, the notion of creating a document by 'tearing off' the top sheet of an appropriate template.

I guess you didn't use OS/2 2.x or Warp much, then?

It had a very similar system. I personally found it a useless distraction. Clever but it did not play nice with also being a superb platform for DOS and Win16 apps.

But then, OS/2 never embraced that enough. Example: if you formatted a drive with HPFS, you got long filenames. But there was no filename munging* to 8.3 names, so DOS and Win16 apps simply could not see any folders with long names, or files with long names.

:Facepalm:

Such an obvious hole but nobody spotted it.

MS's ugly hacks on top of FAT to add LFNs (VFAT) and FAT32 actually worked to its advantage. LFNs were aliases to short names, so DOS apps could see and access the short names. And if you didn't defrag it or CHKDSK it, it was fine and quite robust. You could reboot into DOS and it worked.

Inspired. Excellent highly-useful functionality because it was a hack.

The better design -- HPFS -- had a usefulness cost.

There's some kind of lesson in there.

And then later MS's counterblow with Win32s was very clever and very effective. Win32s wasn't very useful to end-users but it was very handy for developers and it killed WinOS2.

I do not like Microsoft. I do not admire Microsoft. I think they're rather evil, nasty and treacherous. But by the gods, they've done some amazing work.

I liked OS/2 2.0 a lot. I liked Concurrent DOS a lot, and CDOS386 was amazing.

But in an alternate universe where Intel didn't "fix" an erratum in the 80286 and cripple CDOS286, and it throve, we might never have got OS/2 at all. No OS/2, no Windows 3. No Windows 3, no NT. And NT is a good OS and xNix' only real rival.

Or let's say that happened but MS beat IBM's argument and OS/2 1.x was a 386 OS. Then OS/2 would have succeeded and then NT wouldn't have happened.

NT was, I'm sorry to have to say, a better OS than OS/2 2.x.

A world of CDOS+GEM derivatives, or OS/2 derivatives, might have been far clunkier than today.

Or maybe, to compete with an ascendant Digital Research rivalling the continuing IBM+MS alliance, GNU would have gone with a BSD kernel for the GNU OS in 1988, and Linux would never have happened?

Then we'd have DR versus MS/IBM versus GNU/OS?

---

• I believe "munging" actually is the technical term here. I kid you not.

[tpear]

I’ve never tried OS/2 — one of the few not in my collection. A joint effort of MS and IBM; no idea who had most input to it.

I think Apple had to make a choice between the 8-bit ][ stream — even after upgrading to 16-bit in the GS — and a more future-proof 16/32-bit stream of the Mac. So hobbling and then killing the GS was really the only commercially viable route. I’m not aware of questions over yield.

Have only rarely come into contact with Apple ][ helping a friend with his 2c. If I were to get one now as a retro thing, I think it’d be the GS as I just like the look lol

I agree that the WinNT architecture -is- good. Guess that’s what you get by employing VMS's creator.

The whole Amiga/ST thing was a crazy time. Atari initially helping fund the Lorraine, Tramiel being ousted from Commodore & buying Atari, Commodore snatching Lorraine from Atari and the ex CBM people Tramiel took to Atari creating the much simpler ST design. You couldn’t make it up 😊

I have a feeling I once saw Commodore's own 16-bit machine at a show — the Z8000 based unix box. Don’t really remember much apart from seeing it.

More interesting (to me, I mean) than if GNU had adopted BSD is if they’d actually made headway with Hurd… in the end, Linus's clone of Unix won that battle instead so we’re stuck with 1960s design with masses of boot-ons and bodges to try to hide how badly it creeks at the edges from old age.

[liam_on_linux]

I’ve never tried OS/2 — one of the few not in my collection.

It's worth a look. Warp is an interesting OS, and the fact that a current version still survives says a lot, IMHO.

Warp Server 4.52 is out there for nothing, supports SMP and USB, and runs in a VM fine. VirtualBox was built for OS/2 Warp.

A joint effort of MS and IBM; no idea who had most input to it.

OS/2 1.x is kinda sorta 50:50.

OS/2 2.x (and Warp) are 100% IBM.

I think Apple had to make a choice between the 8-bit ][ stream — even after upgrading to 16-bit in the GS — and a more future-proof 16/32-bit stream of the Mac. So hobbling and then killing the GS was really the only commercially viable route. I’m not aware of questions over yield.

Agreed.

Have only rarely come into contact with Apple ][ helping a friend with his 2c. If I were to get one now as a retro thing, I think it’d be the GS as I just like the look lol

Pretty much me too! :-)

I agree that the WinNT architecture -is- good. Guess that’s what you get by employing VMS's creator.

Indeed.

NT was based on OS/2 3, the CPU-independent version, of course.

Classic MacOS is not a miracle of OS design. (UI design, yes.)

But compared to GS/OS, yeah it was streets ahead.

There's an essay in exploring the problems and issues of OS development in the 1990s and in the 21st century.

'90s: aargh, we have to boot native, we have to support 4GB of RAM and 4 different APIs and 2 different buses and it's a nightmare! We've been trying for 2 years but if we move to API 2, it breaks everything that uses API 1, and that includes our whole UI! But now we have API 3 and we haven't even thought about API 4 yet!

'00s: OK, we'll run this on the host and spawn a VM for the old stuff. Give it 64 meg, that'll do. Tweak it so it knows it's in a VM and plays nice.

The whole Amiga/ST thing was a crazy time. Atari initially helping fund the Lorraine, Tramiel being ousted from Commodore & buying Atari, Commodore snatching Lorraine from Atari and the ex CBM people Tramiel took to Atari creating the much simpler ST design. You couldn’t make it up 😊

Quite so!

I have a feeling I once saw Commodore's own 16-bit machine at a show — the Z8000 based unix box. Don’t really remember much apart from seeing it.

Ever read The Secret Weapons of Commodore? So many great ideas. So many terrible ones. And somehow often the terrible ones made it to market and the great ones didn't.

More interesting (to me, I mean) than if GNU had adopted BSD is if they’d actually made headway with Hurd…

That's a fair point. If they'd gone with the BSD kernel, that would almost certainly never have happened. OTOH I think they over-reached themselves dramatically there. Cf. my '90s-vs-'00s paras above.

in the end, Linus's clone of Unix won that battle instead so we’re stuck with 1960s design with masses of boot-ons and bodges to try to hide how badly it creeks at the edges from old age.

Yup. And now Plan 9 is GPL but it's too late and nobody knows or cares. If I had Shuttleworth's money, I'd probably start there...