OS/2 and the path to freedom

IBM won't update it or liberate it. Its fans should find new ways to free the OS/2

The story of OS/2 is sad yet informative. The offspring of an unhappy marriage, the operating system went through a stormy adolescence and a broken home, ending up in the sole custody of IBM after Microsoft ran off with the Windows cleaner. Despite an expensive education and being sent to work in the City, OS/2 is now closeted in semi-retirement.

However, some of its friends want it back in full force. For a second time, the OS2World.com community has petitioned IBM to open source the system; for the second time, quoting unspecified "business, technical and legal reasons", IBM has declined to do any such thing.

That's an exceptionally sane response. There is lots of non-IBM code in OS/2 from Microsoft, Micrographx and others, which simply can't be thrown open. There are lots of security-conscious clients still using the operating system, who would not be at all happy for the inner workings to go on show. There are existing licensing agreements with what remains of the OS/2 reseller commnuity. Picking that lot apart safely would be expensive and lengthy and have no strategic benefit. Quite the opposite — IBM is keen for its existing OS/2 customers to move onto more modern systems.

But insanity has its own rewards. Apart from annoying Microsoft, an open OS/2 would allow the existing users a far more flexible migration path, allowing critical subsystems to be interfaced with Linux or other frameworks. It would help share some of the knowledge behind what is widely regarded as an exceptionally reliable and capable operating system. After all, IBM is in favour of open systems, isn't it?

IBM could open up whatever parts of OS/2 will cause it least pain, and let the community fill in the gaps. Failing that, and given that OS/2 is now utterly stable and defined, the community could just set to work to rewrite the lot from existing documentation: a big job, but probably not as big as shifting a stubborn IBM. That would achieve all the results they want, plus one intangible extra — proof that the real secret of open systems is never having to take no for an answer.