Subject: PC: Classifying Tool (over network, without reboot)
Date: Wed, 13 Jan 1999 20:14:17 -0500
Subject: RE: PC: Testing Tool (reboot)
You are on the right track but going in the wrong direction. The reboot test
is often mistakenly thought to test to see if the machine will reboot post Y2K. Reboot
tests look at the millisecond during a reboot before the system clock kicks in to see if
the PC returns a valid date post Y2K at that precise time. This is an RTC issue and I
really don't care to enter that debate.
I agree that an RTC test (i.e. to see what happens in the first 5 ms after the turn of the
century) is worthless. That said, I must tell you that test is an entirely different thing
than a reboot test.
I don't agree with this definition of a reboot test. A reboot test
determines how the PC will react to the next century the first time it is booted. It
checks the instructions in the BIOS PROM. It is critical, not trivial because every PC
will be booted at some time after the turn of the century. Even if a PC is up 7x24, there
will come a time when it will be taken down for servicing, then it will be booted the
first time after the turn of the century when the power is reapplied.
It is absolutely critical that every PC type be tested with a reboot
test. It is impossible to accomplish without booting the PC because, if you attempt it
without rebooting, you are testing the data structure that holds the date information, not
the binary code (i.e. program instructions) that are executed at boot time. This makes the
critical mistake of assuming that the BIOS program was written to react well to the turn
of the century. The reason we are all doing Y2K work now is that a few program were
written that did not react well to the turn of the century. The BIOS program in your next
computer may be the next problem discovered. That is why we do a reboot test.
Has your new product discovered a way to perform this
test without rebooting?
I don't know if Ed's tool does (I think I just proved it can't), but I
do know my tool has the next best thing - accurate identification of the BIOSs across a
network, even if the manufacturer has tweaked the BIOS, it will distinguish between the
BIOSs (within the tools limitations). This allows you to perform a reboot test of
just one sample PC of every type rather than testing all truly identical PCs just to catch
the ones that are tweaked.
What are the limitations? (1) The classification does not work on NT (because of NT's
security features). (2) Some PCs may be assigned to different categories that are actually
identical PCs (but the opposite will not happen, assign PCs to the same category which are
not truly identical). The process can properly identify these few motherboards that react
this way, but it takes running the test over several reboot cycles (such as installing the
test in the startup folder and collecting the results after a week or so of daily