OS/2 2.0: A house built on rock by Gene Wilburn Copyright 1992 Gene Wilburn [A slightly trimmed version of this article was published in Computing Canada, April 13, 1992, p.13, as "OS/2 2.0 Worth the Wait"] The most curious and bothersome aspect of the PC industry is that it lacks a foundation--a fundamentally sound operating system. It is a house built on sand. MS-DOS continues to bedevil the industry with its 640Kb barrier, lack of process protection, and single-threaded, single-application design. DOS extenders such as DesqView and Windows are kludges--attempts to retrofit DOS with improved functionality. But one TSR conflict, one inappropriately masked interrupt, and the whole edifice collapses. Re- boot time. Users with high-performance PCs deserve something more sophisticated. It is absurd to need "stealth" technology to squeeze an extra kilobyte of device driver into the DOS UMB attic while megabytes of RAM go unused. No one loves DOS. What we love are our DOS and Windows applications. What if we could take all the applications we know and love and start over, on a new operating system? A modern 32-bit operating system with preemptive multi-tasking, multi-threading, demand paging, virtual memory, flat memory addressing, and protected sessions? That is exactly the choice that lies before us. OS/2 2.0 is here, it's real, and it has the potential to fundamentally alter the face of the PC industry. IBM has delivered an advanced PC operating system that is going to force us to re-evaluate what PC computing is, what it should be, and what we want from it. In a move that is nothing short of brilliant, IBM has made OS/2 2.0 a backwardly compatible operating system. While their advertising slogan-- "a better Windows than Windows, a better DOS than DOS"--encapsulates the concept, I prefer IBM's other, more subdued description--"the integration platform". OS/2 2.0 is the only product available that can run software written for Windows 3.X, MS-DOS, OS/2 1.X, and OS/2 2.0. It can run them all at the same time--each in a separate, protected, session. I was skeptical about these claims until I installed OS/2 on an no-name 486 and found myself simultaneously writing in Word for Windows, consulting a DOS-based GrandView outline, downloading a file in ProComm Plus in a DOS background session, checking user logins on a remote Unix system over a TCP/IP network, and periodically inspecting an OS/2 system usage utility. The download never burped. Everything ran smoothly. After several weeks of operating OS/2 on my home and work PCs, I'm impressed by its solidity, usefulness, and design elegance. Experience with OS/2 changes your perceptions about PC computing. With OS/2 you can carve out the kind of computing you and your users want, mixing together any combination of existing DOS and Windows applications. You can gradually add native OS/2 programs and drivers for networking, CD-ROMs, scanners, and multimedia equipment and never again have to worry about stealing critical memory from applications. For large disk drives, OS/2 offers its HPFS (high performance file system) as an alternative to the antiquated DOS FAT file system. HPFS supplies better buffering, a more optimized structure, and freedom from the DOS "8.3" naming convention. HPFS file names can be up to 254 characters long and can include spaces. OS/2 also resides comfortably in a FAT file system, maintaining its backward compatibility with DOS partitions. OS/2's flat, 32-bit memory addressing opens the door to impressive software possibilities--everything from the old to the new. Mainframe COBOL applications can be brought to the desktop. CAD and graphics programs that rival the performance of Unix workstations can be developed for OS/2. The large addressing space is a natural fit for distributed corporate database modules. The multi-threaded architecture will give a strong boost to multimedia applications. From a planning point of view, OS/2 provides stability of direction. Future releases of OS/2 2.X will be ported to multi-processor hardware and to RISC architectures. Security features will be scaled up with each release. The object-oriented user interface, Workplace Shell, provides a better long-term desktop metaphor than Windows. On the downside, OS/2 is not for meek systems. In addition to a 386 or 486 processor, it requires 30Mb of hard disk for a complete installation, plus a minimum of 4Mb RAM. Realistically, make that 8Mb. At least Windows has softened us up for these requirements. In comparison, Word for Windows 2.0 requires 15Mb for a full installation, and it's just a word processor. It is impossible to discuss OS/2 2.0 without discussing Windows 3.1. IBM and Microsoft are both launching massive advertising campaigns to promote the products. Comparing the two is difficult simply because there is no direct comparison. OS/2 can only be directly compared with Windows NT, Microsoft's OS/2 3.0 project, renamed after Microsoft dropped out of the OS/2 2.0 effort. But Windows NT is still in development and will likely not arrive until 1993. Certainly Windows 3.1 has a lot going for it. It's faster than Windows 3.0 and it traps errors more gracefully. Small design changes improve its appearance, and it works better on networks. It is an improved environment that offers some of the features of an advanced operating system. In addition, Windows has the support of numerous vendors. It's a known, if not loved, commodity. The question is whether Windows 3.1 is good enough, and whether we want to wait until next year to see something better. OS/2 offers us the immediate choice of a modern, carefully crafted architecture. We are no longer, by necessity, forced to follow Microsoft's sometimes mercurial changes of direction. IBM should be singled out for praise for its OS/2 2.0 project. It kept its promises to developers. It stayed the course, producing a product that leapfrogs all existing PC operating system technologies. For a long time I considered the Microsoft Windows juggernaut unstoppable. Now that I've tried OS/2 2.0, I'm not as certain. IBM has launched a product to be reckoned with. Microsoft no longer has a monopoly on PC operating systems. IBM's buy-me-and-try-me campaign, with introductory pricing, is based on their confidence that once you experience OS/2 you will discover its advantages for yourself. They're betting you will prefer a house built on rock. [Gene Wilburn is a senior systems analyst at the Royal Ontario Museum, Toronto, Canada. CIS: 72435,732. Internet: gene.wilburn@rose.com]