There's no question that OS/2 has a promising future. The significant word here, however, is future and IBM's OS/2 exhibit at Spring Comdex underscored this fact. Yes, IBM was passing out a book that listed hundreds of announced OS/2 applications from third-party developers. And yes, the OS/2 booth was one of the stars of the show. But of the 100 OS/2 applications that were supposed to be on exhibit (of which only about 50 or so ended up being demonstrated), less than 30 were shipping, and it seems that it will be at least another year or two before the promise of OS/2 is closer to being fulfilled.
Software developers were frank in their explanation of why there is a relative dearth of released products for OS/2. The main reason, say developers, is simply a lack of demand by PC users.
But before user demand for OS2 applications can be stimulated, several pieces of a puzzle must fall into place. For one thing, systems that run the applications most people want require an 80386 CPU, and Intel, which is currently the only source for 80386s, has not been able to produce enough chips at an affordable price to meet current demands. Nor has the company made any moves to second source the 386; instead it is fighting in court with companies such as AMD that want second-source rights. To its credit, Intel has recently opened up several new plants dedicated to manufacturing 80386s. Unfortunately, some of these plants will probably be producing the "p9," a 386 that supposedly will have a 16-bit data bus. (Intel take note: Even GM realized no one wanted a 6-cylinder Corvette.) Because of Intel's strategy, an 80386 costs around $300, about ten times the cost of an 80286.
Second, the amount of memory required to run OS/2 applications seems staggering, at least to those programmers who were weaned on CP/M. If you want just to boot OS/2, you'll need 1.5 Mbytes of RAM; if you want to run an application, you'll need at least 3 Mbytes. Memory, particularly in large quantities, is expensive right now (1-Mbyte SIMM chips are selling for about $500, double what they cost six months ago), and it's unlikely that it will become cheaper over the coming months.
Finally, Microsoft will have to release a 386 version of OS/2. What the current implementation of OS/2 (written for the 80286) provides that DOS does not is multitasking and the ability to handle a large-memory model. What OS/2-386 will provide is support for a virtual 8086 mode (so that you can have multiple 8086 applications running at once, à la Windows/386), demand paging, and 32-bit memory addressing with large data objects. This, along with a graphical interface like Presentation Manager's, is what I want, and I'll bet that's what most other users want too. When this is available and when 386-based PCs are affordable, end-users will clamor for OS/2 and its applications. Microsoft, however, says it won't start releasing OS/2-386 software development kits until mid-1989.
What this means is that DOS will continue to be a dominant force while other operating systems Unix and the Macintosh OS, for instance-will continue to attract more and more followers. Unix advocates in particular are making major advances, with even Microsoft admitting that demand for Xenix soared after the details of OS/2 were finally released last year. In the meantime, the door is still open for alternate operating systems, and it will remain open longer than many of us originally expected.
Jonathan Erickson
editor-in-chief