E-mail

In his Test & Measurement World blog for April 14, 2008, my colleague Martin Rowe suggested that our readers should sign a petition requesting that Microsoft extend support for Windows XP past the published cutoff date of June 30, 2008. Created by InfoWorld, the petition so far has accumulated more than 180,000 signatures but will probably have all the impact of a June bug hitting Microsoft’s windshield.

Having spent the better part of a week resurrecting my PC and reinstalling Windows after a complicated failure cascade involving a motherboard problem that caused a hard-disk crash in a RAID-1 array, accompanied by unexplainable damage to the master hard drive’s Windows system image, I’m not in the mood to adopt Microsoft’s latest offering.

My last contact with a Microsoft operating system’s innards occurred with CP/M, which I vaguely understood and could modify to meet the hardware needs of various systems from the mid-1970s. I had to understand CP/M because I had no choice, but it wasn’t that onerous a process in spite of Microsoft’s opaque documentation. [Editor's Note: Two readers have alerted us to the error in this paragraph, which Brad Thompson acknowledges: CP/M was a Digital Research product.]

Today, my PC provides me with a workspace, a toolkit, and test-instrument control. I shouldn’t have to care about what’s going on inside the operating system nor worry about anything except the work that I’m performing. And I can’t afford to replace certain older software packages that Windows Vista won’t support.

In an ideal world, Microsoft’s next operating system would feature improved (and understandable) error messages, modular design, and detailed diagnostics that locate and facilitate replacement of damaged parts of the operating system. Put another way, the PC and its operating system should resemble a full-featured oscilloscope. If half of a scope’s screen goes blank, you can consult a troubleshooting manual and check various things—connectors, power-supply voltages, and circuit subassemblies—to isolate the failure to a particular subcircuit or component.

Does any Linux version or the current Macintosh operating system appear logically designed and easier to troubleshoot? I doubt it, but I’m experimenting with Linux. In the meantime, I’ll continue to use Windows 2000 and build on my understanding of its failure modes and foibles. But I’m not interested in starting anew with yet another Microsoft operating system.