Re: Ada & .Net (Rotor)

[dennison@telepath.com (Ted Dennison)]

"Juergen Pfeifer" <nospam@nospam.org> wrote in message news:<a90tsf$tms$07$1@news.t-online.com>...
> >> ...
> >> All that results in a much higher stability
> > > of
> > > Win2000 compared to NT4.
> >
> > But it still is not as stable as Solaris or even Linux judging from
> > the number of reebots after I switched boxes on my desk. :-(
> 
> Then you have other problems. I (and collegues) are running W2K
> or WinXP for several month without any reboot (and we have
> load on the systems).

My home Win2k system (an Athlon box with a non-intel chipset on which
I play a lot of games) tends to crash all the time, particularly while
playing DirectX games. My work NT4 system (A PIII system with an Intel
chipset on which I do development) generally stays up until we take a
power hit at work (or I stupidly test out one of the device drivers
I'm creating on it).

The point of this is:

  o  *Of*course* stability has everything to do with what you do with
the system. It'd have to be a truly crappy system to crash when
nothing's running. :-)

  o  As an end-user, the distinction doesn't really matter to me. The
fact of the matter is that, for whatever reason, I currently find my
Win2k box significantly less stable than my NT4 box. Perhaps they did
clean up a kernel bug or two in NT, but whatever crashes they
prevented are being swamped by the overall buggyness of their hardware
support and their DirectX interface.

Now in defence of Windows, often the real culprit is the hardware.
Over the years I've found that while putting together a PC system is
fairly easy, putting together a *stable* one is pretty hard. Sun has
only a very few hardware combinations to worry about, so its much
easier for them to put together a stable system.

For instance, I think I increased the stability of my home system by
about %300 by just installing a case fan near the CPU. SETI@Home isn't
giving my CPU a breather *ever*, and that takes a toll in heat.
Similarly my wife's Win98 system for a long time had trouble with
reboots whenever her desk was bumped. After a while I got sick of her
yelling at the kids for bumping the desk, and determined to fix the
damn thing. It turned out that the power supply's circuit board had
broken near the mounting screw holes, and was thus swinging freely.
Whenever the system took a big bump, the PS circuit board would swing
back, and the circuits underneath would make contact with its metal
enclosure. Out go the lights.

But still we can't entirely let Microsoft off the hook here. They seem
to get things working fine with the most common hardware, and then
just quit. Folks with second-tier vendor stuff like Athlons and VIA
chipsets are just left swinging. What you end up with is a thousand
hardware vendors (who generally operate in very low-margin businesses)
trying to make their drivers work with 4 Windows OSes and the drivers
of a thousand other hardware vendors, which are all secret code and
moving targets. It just *cannot* be done. I have to constantly scan
the VIA and Athlon and nVidia websites for news of new compatability
problem fixes. The fact of the matter is that users are dying to help
fix problems, but The System in Windows won't let them.

For example, there was one problem some folks were having with crashes
while trying to burn music CD's. Over the course of a month assorted
users narrowed it down to systems with NT and a certian VIA
southbridge chip. VIA denied there was a problem. Next they figured
out that it happened whenever there was activity on both IDE busses,
and the soundcard was just using enough PCI resources to make the
problem show up more often. Then somone managed to reverse-engineer
the VIA bios and find a BIOS hack that made IDE behave diffently and
fix the issue. Later semone else made an NT device driver to apply it
at boot time. About a month later, VIA released a driver with the fix,
so the custom device driver was no longer nessecary. The punchline is
that it turned out that Linux folks had identified a similar problem
and fixed it months before.

I really don't see how the system Microsoft has set up of thousands of
interoperating secret device drivers in a secret kernel can ever hope
to compete for stability with the Linux system of drivers (and the
kernel) being standard and open and available for debugging by anyone
who happens to notice a problem with their particular hardware combo,
vendor or not.

-- 
T.E.D.
Home     -  mailto:dennison@telepath.com (Yahoo: Ted_Dennison)
Homepage -  http://www.telepath.com/dennison/Ted/TED.html