Writing Sample:
Street Talk column from the early '90s in NEWS/3X-400 Magazine

Everything's better when it sits on a RISC, even an AS/400 application. It's been confirmed that IBM has the RS/6000's RISC chip running on the AS/400 in the lab, with MI running on the chip. AIX is also running on the AS/400. We haven't heard anything about how close the "RISC/400" is to being commercially available.

Baby/6000. Look for the September or October announcement of a new low-end RS/6000, the Model 230. The 230 will have a starting price of $5,000.

Mac vs AS/400. For those looking for an alternative to 1960's programming languages, 1970's architecture, and 1999's prices, you may be interested in Apple's upcoming announcement of the Mac tower. Based on a 68040, 25 Mhz processor, the Mac tower will have a main memory capacity of 256 Mbytes running under up to 1 Gbytes of virtual memory.

What makes the Mac tower a competitor to the IBM midrange is that it's a multiuser Macintosh! Applications written for the Mac tower will not only be able to handle multiple users, they will be able to spin off tasks to run on any unused attached Macs. The Mac tower faster processor will allow it be an "application server", where Mac programs appear to run on the attached Mac, but actually run on the faster Mac tower. This means that a Mac Classic (list price around $900) attached to a Mac tower via Ethernet can run programs faster than a stand-alone Mac IIfx!

Compatibility with existing Macs shouldn't be a problem with the Mac tower, as it runs a newer version of the Mac's System 7 operating system. No word yet on pricing (or the official name) for the new machine.

Speaking of IBM and Apple, we've noticed some fairly poor reporting and analysis of the joint Apple/IBM venture in the midrange press recently. More accurate analysis and reporting can be found in MacWEEK, which is free to qualified applicants (or for $99/year) from MacWEEK, c/o JCI, PO Box 1766, Riverton, NJ, 08077-7366, 609/461-2100.

At last, the promise of a faster MAPICS. According to Rochester IBMers, the AS/400's speed can, with current technology, be increased from 10 MIPS to 200 MIPS. On the same note, IBM Austin is confident that its RISC chip will be able to run 1,000 times faster than it does now.

When is an AS/400 not an AS/400? Why, when it's a S/36. So when IBM struts their figure of 100,000 AS/400s sold so far, remember that this includes AS/Entry and Model Y10 sales. We've heard that IBM may have sold as many as 33,000 AS/Entry and Y10 machines, making the real AS/400 figure closer to 67,000.

The Worst Reason to Buy an AS/400. A recent Newslink message from a S/36 user asked how to read the workstation cursor position from an RPG program. He said he had already asked his IBM SE how this was done. Did the SE suggest getting the assembler subroutine from a dozen sources, from third party vendors to published articles in magazines (including this one)? Nope. His solution was "Buy an AS/400."

Take a lesson from IBM: the next time you don't have enough room in your car to take your friends to a restaurant, well, just buy a Greyhound bus.

More "vocabluary." Dennis Cole of James Longley & Co., Ltd. submitted the following vocabluary term from a magazine article describing IBM's Hursley Park facilities in the U.K.: "Another quirk of IBM was that the building should provide a 'minimum of 80% outside awareness' - translated into everyday language this means that everyone should have a window."

Perhaps we can look forward to a Microsoft version of "Outside Awareness 3.0" for the U.K.

Name culpability. IBM is promising POSIX compliance for the AS/400 (POSIX is an IEEE standard for UNIX operating system interfaces). The U.S. government requires POSIX compatibility, hence IBM's interest. However, since the AS/400's architecture won't isn't able to implement many of POSIX's requirements, it will be partial compatibility.

This reminds some industry observers of IBM Rochester's "compliance" with other standards, like TCP/IP, which is at best a partial implementation. We think we've found the third level of compatibility in computing, behind hardware and software compatibility: name compatibility.