As discussed in the entry for Pink, Copland was step one of Apple Computer's contingency plan when Pink fell several years late. The OS was announced in March of 1994.
As compared with then-current releases of the Mac OS (System 7.1), Copland would require fewer system resources, crash less often, remove hardware dependencies, and introduce a rudimentary form of preemptive multitasking.
At the time, Apple was thrashing around with the question of how to increase their profit margins, and licensing the Mac OS was a recurring part of the plan. Therefore, they planned Copland to include a hardware abstraction layer (HAL), which would make it easier for the OS to work on machines that used different low-level hardware than Apple's did. The HAL would act as an interpreter between the OS and the computer's components, limiting developers' abilities to tie a function directly into a specific component. The press at the time blamed the difficulties of implementing the HAL for Copland's delay from 1995 to mid-1996.
My description of Copland's new memory management scheme and intended implementation of preemptive multitasking will reveal a lot about the weaknesses of the Mac OS previously. Under Copland, the Mac OS would work in a protected space, while all programs shared another space. In theory, applications could crash each other, but they could not bring down the OS. A Copland-aware program could spawn subprograms that would run in their own protected space. (Why not have the OS run a stub application that would put every legacy Mac app into a separate protected space? I have no idea, and as Copland never saw the light of day, the question is academic.)
Full preemptive multitasking was due in Gershwin, the next release of the MacOS, a year or two after Copland's release.
Much of Copland was written in native PowerPC code. This was quicker than the 680x0 code that had made up earlier releases, though the conversion to PPC would not be complete until later.
With Copland, Apple planned to open up a large number of APIs that had previously been closed to third party developers, and to extend QuickDraw GX printing (later abandoned) and PowerTalk communications (later abandoned).
Finally, the OS had a couple of nice bells and whistles for users. It had a much more active set of 'help' features, and it would allow greater customization of the user interface (think themes). The Copland UI was shown to the public in June of 1995, to generally positive reviews. Ironically, the UI customization tools were one of the few parts of the operating system ever implemented in a released OS, but they had been disabled because Apple's attack lawyers thought that they reduced Apple's ability to sue people for copying their look and feel.
In November 1995, the Copland beta was shipped to 50 key developers and the release of the OS was delayed until 1997.
In May 1996, a developer release of Copland was promised by summer. Later in the month, new CEO Gil Amelio said Apple would ship Copland piecemeal.
And finally, in August, Chief Technology Officer Ellen Hancock killed the project, icing both Copland and Gershwin. (Mac users note that the company was shipping new machines with System 7.5.3 at the time).
Through the end of that year, Apple negotiated to buy Be, but deadlocked over the price and purchased NeXT instead in January 1997. This established NeXTSTEP as the foundation for future Mac operating systems.
Apple's attempts to replace the aging MacOS with something better were ugly. It was against all odds that they finally succeeded by introducing the unix-based OS X in March 2001.
To be fair, they weren't the only ones thrashing around at the time. Microsoft Bob had been introduced to great fanfare in early 1995, and failed utterly. The company characterized Bob as a 'productivity tool', but it really was its own OS, sitting on top of Windows 3.1. It had its own (proprietary) APIs, and applications written for Bob could not run without it.
Why did Copland fail? The problem was not, as with Bob, a marketing one. The public didn't get a chance to reject it; it was never released. Nor, I suspect, can we blame the developers. Parallels for everything promised in Copland had been delivered elsewhere. I suspect that Apple's project managers were weak, and that the head of the OS group was subject to evolving demands from the CEO and marketing groups. This would have made the release plan fluid, delaying builds and requiring programmers to change their focus mid-project, wasting momentum and causing newly developed code to be thrown out before it was ever used. The periodic loss of key employees could not have helped.
When NeXT entered the picture, this changed. Avie Tevanian, VP of Engineering at NeXT, and soon Senior VP of Software Engineering at Apple, had been the principal designer of the Mach operating system and of NeXTSTEP. He had the authority to reject demands from marketing and Steve Jobs, who soon became CEO, trusted him enough to stay out of his way.
Byte Nov 1994
Macworld April 1995