Gil Amelio's 500 Days at Apple

Sabah Arif writes "Apple Computer was teetering on the edge of bankruptcy on January 31, 1996, when Gil Amelio succeeded Michael Spindler as CEO. The first thing he did was turn down an acquisition offer from Sun Microsystems, then he moved to secure Apple's short term financial future by having a huge bond sale. As he restructured the company (and cut 3,000 jobs), Amelio realized that the Copland project would never finish, and decided to buy NeXT Software, paving the way for Steve Jobs' triumphant return in 1997. Read the whole story of Amelio's 500 days with Apple."

Comments

Re:I used Copland

[squiggleslash]

If you used a copy, you used the developer-only pre-release because a finished version of Copland was never actually released to the public. It was cancelled largely because:

1. In terms of design, it was never the next generation operating system. It was supposed to be a stepping-stone to a future OS (codenamed Gershwin, which allegedly never had a single line of code written for it.)

   Aside from anything else, memory protection and pre-emptive multitasking was only provided for applications that weren't visible to the user (ie all interactive apps had to sit in the same memory space and they had to cooperatively multitask.) You could, obviously, structure your app with an interactive stub that communicated with a protected program that contained the meat of your application, but that's convoluted, and still puts the user at risk of one bad piece of code upsetting the rest of the operating system.

   It's questionable that this "design" even constituted a stepping stone, and if it does, it's only to an OS/2-like "Next generation OS" - pre-emptive multitasking and memory protection for processes only, not the multiuser-style security that we see in NT and Unix. Copland would have been a stepping stone to an operating system competitive with Windows 95, with everyone at Apple aware that Windows NT was already on sale and being touted as the eventual successor to anything Microsoft did with the original DOS/Windows codebase. In October of 1995, after Windows 95 had been released, the target date for Copland (not Gershwin) was moved to "some time in 1997". You can imagine the panic that must have occurred at Apple at this point - this meant Apple wouldn't have something "as good as" Windows 95 until sometime well after 1997.

2. The requirements list was changing daily, and the OS was taking longer and longer to produce. One infamous legend has it that Amelio did a presentation on the OS and was immediately slapped down by developers who wanted to know why it didn't support multi-threading. Amelio, who possibly had no idea what any of this actually meant, promptly announced that it was going to support multi=threading. Copland's spec was so crude that adding multithreading was seen as essentially something that could only be done if that part of the OS hadn't been written yet - eg it necessitated a rewrite.
3. As you yourself saw, it was extremely unstable in '95. Allegedly, cleaning the code base up so it just worked was taking too long by itself.

The fact that Copland just plain wasn't the next generation OS it was being presented as is the unwritten reason why, I suspect, it was eventually cancelled. Users who got it would have ended up with an environment just as unstable in practice as Mac OS 7. Rogue apps would have still crashed the entire system, as developers would have almost certainly ignored the requests to break up their application into back-end and front-end code, because there was no real incentive for them to do so. It would take real signs that Gershwin was a real OS before anything moved forward.

I reread the specs recently, and I have to admit, I wondered what the hell Apple had thought it was doing. Gershwin should have been Apple's focus from the start. I assume they thought they were under pressure to release something that showed progress, but I doubt Copland would have done anything to help Apple's credibility, and as it was, it did a lot to harm it, as part of a pattern that lasted until, perhaps, the last three or four years, where developers were told one thing, and then promptly told to ignore it.

Google for QuickTime source code lawsuit

[metamatic]

Apple had discovered that Microsoft had stolen QuickTime source code and used it in Video for Windows. The trial was going very well for Apple. Yet it was suddenly dropped when Microsoft agreed to make the "investment" in Apple.

Ask yourself this: if the "investment" wasn't under threat, why do you think the full terms were kept so secret?

Re:I used Copland

[argent]

One thing to keep in mind is that Apple would have been targeting 8MB machines in the mid-90s and probably could not have afforded the overhead of a Classic VM.

Apple was running a classic virtual machine under System V UNIX in the *early* nineties. But classic Mac OS was born doomed, the API guaranteed that nobody would ever be able to do multitasking under Mac OS without using fixed partitions... which was a performance killer for low-memory systems even with demand paging. They should have replaced the API by 1990 with one that used opaque handles like UNIX, or required explicit locking of handles during use. That would have allowed a single classic application alongside multiple New API applications, which would have been good enough for a transition if it had been started early enough.

Given that the classic environment in A/UX was System 6, they were actually on the way there. But System 7 incorporated The Grand Maltitasking Charade by default and they couldn't really go back after that.