__/ [Roy Culley] on Sunday 25 December 2005 01:03 \__
> In a recent so-called Desktop Linux versus Windows XP shootout,
> writer George Ou declares that "Microsoft handily beat the open
> source platform." The basis for this judgment? "OS boot time and
> application load times" on two different PCs. Oh, dear. This isn't
> right at all.
> First, using OS boot times as the only real comparison between
> operating systems is a lot like comparing cars by how long it
> takes you to get from 0 to 60. Yes, it's a measurement, but by
> itself it doesn't say much of anything.
> The a very nature of the test itself is really pretty meaningless,
> anyway. Two machines do not a benchmark make.
> I used to help write and run benchmarks at Ziff Davis Labs, now a
> branch of Veritest. The very idea of saying anything about
> operating system performance based on two machines load time
> wouldn't be taken as anything but a bad joke.
> That said, Office does make good use of Windows to load
> quickly. I, however, don't see this as a good thing at all.
> Office loads quickly, because it -- like far too many Microsoft
> programs -- makes use of Microsoft's built-in integration between
> its applications and its operating systems using shared libraries,
> and its eternal confusion between data and programming code found
> in everything from DDE to OCX to ActiveX.
> This is Microsoft's eternal Achilles' heel. By integrating
> applications and operating systems at a deep level, Microsoft
> assures that any security hole has the potential to have profound
> effects on the entire system.
> As it happens, Ou and I talked about this very point a few weeks
> back. His response?
> "A lot of those issues have been resolved. IE on SP2 is very
> protective about ActiveX, and Vista will be even more protective."
> Yes, Windows is better than it used to be, but the fundamental
> flaws that come from an operating system that was never, ever
> meant to work on a network are still there.
A "Linux versus Windows XP shootout" over Christmas? Slow down, cowboy. It's
no news either: http://blogs.zdnet.com/Murphy/index.php?p=459
"What's noteworthy about it is that Microsoft compared Singularity to
FreeBSD and Linux as well as Windows/XP - and almost every result shows
Windows losing to the two Unix variants. For example, they show the number
of CPU cycles needed to create and start a process as 1,032,000 for
FreeBSD, 719,000 for Linux, and 5,376,000 for Windows/XP."