W

ARPSTOCK --- The final time slot of the day was committed to the summary sessions, and I chose Lynn Maxson's "A Call To Action."  I believe I selected well.

           Mr. Maxson has put together an eye-opening proposal on keeping your software investment intact no matter what hardware platform you choose or migrate to.

           First, a little full-disclosure here.  Lynn Maxson is a friend of mine.  He's rowdy, opinionated and occasionally short-tempered.  But he is one of the most brilliant people I've ever met.  And he is presenting a classically successful scenario to liven up the future of OS/2.

           Lynn's proposal is this:  create a $20-per-year international membership organization that will create, hold and disseminate the source "code" for the OS/2 operating system and the apps that go with it.

           Mr. Maxson began his lecture by handing out a far-reaching, well-thought-out and certainly complete four page summary of his plan for the future, and then led the ensuing discussion in a somewhat happily-raucous session among the 50 or so attending Warpstockians.

           From the specification of machine hardware to the generation of machine code from program specifications, the Maxson Proposal has some eye-opening points.  For starters, this isn't a volunteer organization.  There are paid full-time positions to keep the organization on track.  Moreover, programmers don't start coding when they only have most of the specifications in place (a common "R&D" technique).  In fact, they don't start coding even after they have all of the specifications in place.  That's because current compiler technology is to be extended to the specification level, and the input to the "compiler", or "logic engine", is the set of specifications itself.  The logic engine assimilates the specifications, logically enumerates the possibilities, and determines the absolute best machine code for the original problem.  The first "problem" is to be OS/2.

           Lynn's proposal is a good one, and it can succeed for several reasons.  First, most major software projects (such as the OS/2 kernel) are developed by a handful of people, not hundreds, so the proposal is doable.  The OS/2 kernel isn't monstrous in size, and many of us have coded similar-sized projects on our own.  There are already plenty of software tools that can be used to look at the existing kernel and create a set of specifications from it.  The "logic engine" which Lynn is promoting already exists and works.

           An alternate OS/2 kernel, owned by its users, removes the "kernel fear" that keeps us awake at night.  There would be an OS/2 version for Merced, and for McKinley after that.  There would be a version for any processor you cared to specify.  The compiler optimization technology to do this already exists.

           I hope Lynn takes his Maxson Proposal further and starts to implement it.  Linus Torvalds and Sander van Leeuwen have demonstrated what a few dedicated visionaries can do.  OS/2 deserves the same.

           There is no reason an alternate OS/2 kernel cannot be written.  The Maxson Proposal, or something similar, can succeed.  The idea of a new non-IBM kernel would instantly put OS/2 in the top tier of operating system options for many currently-reticent corporations.

           I hope, I really hope, to see it soon.

           For other Warpstock '98 articles see the Warpstock '98 Article Index.

Lynn Maxson, lmaxson@ibm.net