🚀 go-pugleaf

Ian Staples <ianst@refer.to.sig.au> wrote in message news:4246475F4AE6D211808D006097A1716402E71ECA@aricrmntnews.dpi.qld.gov.au...
> Rod Speed <rod.speed@usa.net> wrote

>> Unfortunately for you, it can be a symptom that your hardware
>> has a problem. A Win98 install is working the system rather
>> harder than most people use their PCs, so it can show up hardware
>> problems that have been lurking since before the reinstall.

> I've often seen this given as an explanation why things fall over during
> an install of Windows [going back to 3.0].  While it may well be true,

It is true, and I've proven it to be true plenty of times.

And carefully tested what it is about the hardware thats a problem
too. On one of most recent occasions its was motherboard not liking
a mix of parity and non parity memory, but ONLY when attempting
a Win98 on a pure SCSI config harddrive and cdrom wise.

On another, both copys of an ST3660A I happen to have
lying around produce it, in a wide variety of motherboard/cpus.

> to the uninitiated it does have a ring of "shark infested seas"
> and "a lucrative new contract" as the tabloids invariably
> phrase these things to fill up a column-inch or two.

Its not hard to prove it conclusively if you have enough spare
hardware around to allow you to work out what bit of hardware
is the problem. And the time to do it. Can be a bit time consuming,
tho most of the time just goes letting Win9x go thru the install.

> So, just what *is* Windows doing that stresses everything so much?

Its just doing a lot more work than most do on their PCs when its
installing. It is after all writing a lot more to the hard drive alone than
most do in months on a particular PC, all lumped into about an hour
with the install. Its also rebooting a few times with a fairly minimal
set of drivers etc, so can got flat on its face more easily.

There is also plenty of obvious potential for not adequately
identifying all the hardware properly with a clean install.

> Is it just that it tries *everything* in the
> process, so any weakness at all will show up?

Thats certainly part of it, particularly with hardware detection.

> As a corollary, I assume the same weakness would
> cause problems if you ever got round to actually
> needing that "bit" in the normal course of work?

Yeah. In the case of the system which didnt like mixed memory,
the problem was just that the way it used memory in the later part
of the install caused one of the boots to just continually recycle.
It would reboot as it was supposed to, then when running himem
there were memory errors which werent seen when it was running
in protected mode as it would normally do after fully installed. So
the auto detection of being partially installed after it had booted
with the scsi drivers in use caused to go flat on its face, reboot
when it wasnt supposed to, and trigger the auto detection of
partly installed when rebooting. Endless reboots until turned off.

I havent gotten around to looking more closely at what hits the
fan with the ST3660A drives. The symptom is dead obvious,
one dud subdirectory off the root with crap entrys in it.

> (That is, it's not the "stress" so much as simply a
> one-off referencing of a bad part of the system.)

Its clearly stressing the system much more with respect
to the hard drive in an hour than most do in months. In other
situations it can be just because the system is used in real
mode for part of the install rather than the protected mode
most Win9x systems spend the vast bulk of their time in.