Re: Grub Failure when HDD descriptor changes

[Yoshinori K. Okuji]

On Tuesday 22 March 2005 09:35 pm, Molle Bestefich wrote:
> Yoshinori K. Okuji wrote:
> > As grub-install says, it is just a guess. In reality, it works in 99.99%
> > machines.
>
> If this was Japan, I'd probably have my head chopped of for being
> insolent by now.

I am not in Japan. :p

The reason why it works in 99.99% is very simple: Even if the device mapping 
is wrong, GRUB boots up correctly if it is installed into the same drive as 
stage2. This is because GRUB receives a drive number dynamically at boot 
time, if it is the same drive. And, the guessing process does a good job as 
long as the user does not change a BIOS setting and have usual devices (such 
as IDE disks).

> But, uhm, is that not just a number that you've pulled out of your hat?

Both yes and no. If GRUB works in most environments, in particular, in 
non-technical users' environments, I don't feel worried too much. This may 
not mean that I don't care at all, though.

> It's practically never worked for me (as in "ever!"), on a number of
> different machines.
> Goes both for boxes where I've setup the hardware and boxes where I
> haven't.
>
> I have one machine where it's worked, and I've tried on at least 7.

You are very unlucky... What kind of hardware have you ever used? Always HPT 
devices?

Okuji