Apple's decision to delay the release of MacOS X - the long-awaited successor to the Mac's venerable system software - disappointed many Mac users, but came as no surprise. The story of Apple's attempt to create a brand new, up-to-date operating system has had such a soap opera quality that as the final episode nears, a last-minute plot twist seemed inevitable.

You almost expect the chief executive officer Steve Jobs to lean out of the shower and tell us all it was just a bad dream, Sue-Ellen.

And to Mac users, MacOS X - it's a roman numeral, not a letter - is something of a dream. It promises Mac owners the kind of crash-free computing they've hoped to get from their system software but have never had. And it comes with a new graphical user interface, Aqua, that Apple claims will put the Mac as far ahead of Windows as it was back before Microsoft released Windows 95.

Sounds too good to be true? It's beginning to feel that way too. Last January, at the bi-annual gathering of Mac fans, MacWorld Expo San Francisco, Jobs unveiled MacOS X and promised to ship it, shrinkwrapped "in the summer" and bundle it with all shipping Macs early 2001.

But at last month's Worldwide Developers' Conference (WWDC), we learned the new operating system, built pretty much fresh from the ground up, would arrive early next year, with the summer's release being relegated to a public test release of the as yet unfinished software - a "beta" version. Apple's spin doctors immediately attempted to persuade software authors at the conference and Mac users reading reports on the internet that the six-month slip was nothing of the kind. "We're delivering the same software at the same time, but with different names," was how Phil Schiller, the company's VP for product marketing, put it. Which begs the question: how come you were previously willing to ask users to pay for unfinished software, but you're not now?

Now the rotational paramedics go quiet - surprise, surprise. May's six-month setback is just the latest in a very long line of broken promises. Understandably broken, to be fair, since the creation of an operating system - the code that governs how a computer looks, feels and behaves - is no simple task. Even the mighty Microsoft regularly put back the release of what became Windows 2000 but started out as Windows NT version 5. And Linux has taken the best part of nine years and thousands of net-connected collaborators to reach the stage where it disturbs Bill Gates's sleep.

In comparison, the genesis of Apple's next-generation operating system doesn't seem too extraordinary. But then we're not talking about the gestation of a single product. Unlike Linux and Windows 2000, MacOS X is the latest in a line of would-be successors to the classic MacOS, stretching right back to a 1987 research project called Pink.

Pink ended up being spun out of the company as part of Apple's bizarre liaison with IBM in 1990, sent out into the world as Taligent, only to col lapse, product unshipped, in 1995. The irony: Pink/Taligent was an attempt to build an object-oriented OS - one made of small, highly efficient and re-useable chunks of soft ware code - something that Jobs, by then out of Apple, was hard a work on at his new company, NeXT. After Pink, Apple engineers in 1993 initiated project Raptor, the company's second attempt to build MacOS: The Next Generation. Raptor pretty much failed too, being finally rolled the following year into try three, Copland.

This time Apple bosses figured they had the right strategy: create a brand new Mac operating system and imbue it with the features that would beat Microsoft's MacOS killer project, codenamed Chicago.

Copland was originally codenamed Capone, after the mobster who brought the city of Chicago to its knees. But the release of Chicago (the OS) in 1995 as Windows 95, while Capone was still in Apple's labs, forced a hasty name change. It also brought a wake-up call to the Mac maker, which suddenly released it wasn't the main man any longer. In early 1996, Apple officially announced Copland and promised to ship it around the middle of the year.

By then Apple's management was imploding, and Gil Amelio, brought in to rescue the company, had seen that Copland would not only be incompatible with existing Mac software, but wouldn't provide users' applications with any of the new, more powerful OS benefits. He canned the project and decided to buy in its replacement. Eventually, Apple bought NeXT, Steve Jobs' company, you'll recall, just before Christmas for $400m.

The following month Apple announced Copland's replacement, Rhapsody (next-gen operating system number four), and pledged to ship it mid to late 1998. Unlike Copland, Rhapsody would provide that all-important backwards compatibility and bring in a powerful object-oriented approach to application creation, plus the two key features: memory protection, which allows an application to die without bringing the entire system down (as anyone who's experienced an Unexpected Type 1 Error will appreciate), and pre-emptive multi-tasking: the ability to run applications simultaneously and give each a full share of the host computer's hardware resources.

Both sound highly abstruse but are essential to creating highly stable systems that don't fall over every time QuarkXPress goes off in a huff. And that was central to Apple's Rhapsody strategy: to provide its professional users, and those who develop software for them, with a solid operating system that could compete with Microsoft.

For once, Apple was able to ship Rhapsody broadly on time. As 1998 rolled round, the company released MacOS X Server. Of course, in "Operating System Street", nothing is quite that simple, and Apple had already been forced to rethink Rhapsody/MacOS X by its major developers, led by Adobe, the creator of PhotoShop, who didn't want to rewrite all their popular applications solely to take advantage of the new OS features.

So while Apple was spinning MacOS X Server as a kind of preview to much-promised next-generation OS, it was reworking MacOS X proper to allow existing Mac applications to use memory protection and pre-emptive multi-tasking, using a technology called Carbon. Carbon is essentially a re-write of the classic MacOS software toolbox that fits it into the new OS's modern framework, allowing modified apps to operate as if they had been designed for MacOS X.

Alongside Carbon sits Classic, formerly known as Blue Box, the MacOS 9 emulation module, and Cocoa, the object-oriented application framework derived from NeXT's bought-in OS and the basis for next-generation applications. All provide three ways of to run applications on a MacOS X-based Macintosh, maximising the system software's backwards compatibility and making the most of its new features. Add the new OS's powerful Adobe Acrobat-based 2D graphics system and the industry standard OpenGL 3D graphics technology, advanced networking and internet access, and Apple's well-established QuickTime multimedia technology, and MacOS X is shaping up to be a very powerful operating system indeed, one that, unlike the classic MacOS, can compete with Windows and Linux.

For all Steve Jobs's frequent comments that Apple has "lost the OS war", this is really what MacOS X is about. That statement is really about how Apple isn't about to become a software company the way Microsoft is - at least until the US Department of Justice breaks it apart - but a system business. The Macintosh is as much a hardware platform as an operating system, and Apple wants the two to fit snugly together. How else is it going to sell systems that are more expensive than their Wintel equivalents?

T he snag here is that MacOS X needs to fit two markets: the professional buyer and the new, iMac-hungry consumer. Classic MacOS works well in this joint role, providing consumers with a user-friendly user interface and the pros with a stack of well-established applications. Unfortunately, MacOS X, while providing the stability professional users crave, hasn't yet proved itself on the applications front.

Apple representatives mention the 200-odd developers who are creating MacOS X applications - almost all of them simple Carbonisations of existing programs - but are unable to say when we'll see any but a handful of them. That no doubt explains the latest delay. "We'll be shipping a final [Mac OS X] 1.0 with pre-loading options in January," said Jobs at WWDC, a far cry from his MacWorld statement: "Mac OS X will go on sale as a shrink-wrapped software product this summer, and will be pre-loaded as the standard operating system on all Macintosh computers beginning in early 2001."

And "there is certainly more work to finish on the details of [Mac OS X]. There is some fine-tuning, but the majority of the product is in place", says Phil Schiller. Apple probably wants to give itself more time to make MacOS X's decidedly NeXT-like user interface, Aqua, appear more like the classic Mac. Again, it's a sign of the OS's schizophrenia: a cute, gizmo-filled front-end to tease the consumers, but one that does little for Apple's professional hardcore. If the reactions of pro users I spoke to at MacWorld are anything to go by, they don't care for all Aqua's UI flim-flam. Can MacOS X become the genuine successor to the classic MacOS, a "one size fits all" operating system for pros and consumers alike? Until we get to see the public beta, with its complete feature set, it's difficult to say.

Certainly, Microsoft's attempts to merge Windows 2000 and Windows 98 into a single, core OS haven't gone according to plan, and aren't expected to deliver until after the successor to 98, Windows Millennium, is released. Even Apple seems concerned - hence the OS's relegation to "optional" status. But Apple's need to get its OS strategy right is greater than Microsoft's and that could ensure MacOS X has the cross-community appeal it needs. The classic Mac OS is so far behind the times developers are already turning away from it - even the upcoming next-gen Amiga OS looks more advanced. And after so many false starts - Pink, Raptor, Copland and Rhapsody - Apple needs to prove it can do the business. And if that takes a six-month set-back, so be it.

• Tony Smith is managing editor of The Register, the IT newssite with attitude

• For more on Mac OS X see: www.apple.com/macosx