Windows is on every desktop. You should put all your eggs in one basket and develop only Windows software, right? Is that really the case? Has the whole industry effectively become so monopolistic that any other choice is a mistake?
These days it is rare for a developer to even look at other operating systems. Like a horse is fitted with blinkers to avoid being frightened and attracted by all what surrounds it, developers are pushed by commercial pressures not to even try any alternatives to Windows.
Weird situation since alternatives fill the past of anyone who has spent more than a couple of years developing software. Some started on BBC computers, some on Sinclair ZX80 or Spectrum, some on Apple II... not to forget mini and mainframes. If you had blinkers at that time you would never be using Windows today.
Even Microsoft managed to look outside the restricted view of their blinkers. Back when it was involved in OS/2 with IBM and decided to divide the world between them, it was advocating OS/2 as the only OS developers should use whatever the target OS was. The rest is history.
History is filled with now successful products which took years to enter the field of vision of most developers.
Nobody predicted the Web though Ted Nelson started his hypertext crusade back in the mid seventies.
Nobody predicted Java though bytecode interpreters, object-oriented languages and garbage collection have been used for more than fifteen years.
Nobody predicted symmetric multi-processing on the desktop... the list goes on.
What is happening today that will fill this list tomorrow? I urge you to take your blinkers off and look at alternatives. Today, I decided to take off mine and buy a BeBox: a computer which can only run BeOS (and SMP Linux). BeOS has attracted a few thousands of developers and no user so far, definitely not in the same league as Windows.
BeOS is a small OS, the kernel and associated tools take less than 10 MB, less than the size of a Windows application. It is SMP from the start, fully multi-threaded and comes with an integrated database. Ok, Windows is multi-threaded too but can you move a window displaying a movie without stopping anything on the desktop? And the registry is more a mess than anything else.
So why did I suddenly went for BeOS? For two reasons: first I was impressed by its speed, architecture... and really cool graphical demos. Second, it's got the same kind of elements which made the Web or Java possible: most of its technology is not new but it's the first time all of it has been integrated in a lean software available on the desktop.
What is the future of BeOS? Hard to say. The fact that Microsoft has abandoned support for NT on the PowerPC might be a boon for BeOS unless of course it kills the PowerPC altogether. Be is rumoured to have started work on an Intel version but, even if there's any truth in that rumour, it won't be available this year.
What should be available to developers in May is the first architecturally frozen version of BeOS called Developer Release 9, or BeOS preview. It will be freely available to anyone in June. The cost of entry is a PowerPC machine and some are reaching the mythical $1000 limit. The BeBox itself is not produced anymore and is slightly more expensive. Talking about collectors items, last year I won an original IBM PC Jr joystick in its original box with unopened documentation. Does anyone wants to get rid of an IBM PC Jr?
(C)1997, Centaur Communications Ltd. Reproduced with the kind permission of EXE Magazine.