Grub missing

[Mario Vukelic]

On Fri, 2007-10-26 at 14:57 -0400, Felix Miata wrote:
> Actually that's a common conception rooted in Linux installers' standard
> non-acceptance of standard MBR code - essentially the same problem Linux
> people complain about. There's no need in most cases for Grub to be in the
> MBR. When Linux is installed to accept use of standard MBR code, there can be
> no damage to Grub or Linux. The only problem with what doz "does to Linux" is
> Linux's own fault for changing the MBR code in the first place.

Thanks for the explanation, it was very interesting. But I do not
understand one thing: isn't it the case that Windows overwrites the MBR
itself, so in most cases there is no "standard MBR code" available at
the time Linux is installed?

I frequently install grub in non-MBR locations on computers that have
Windows installed, and I have played with the "active" flag. It seems to
me that it never booted Linux just by changing the active partition, but
I could be wrong of course.

> Standard MBR code passes control to the "active" (or "startable" or
> "bootable", 0x80 in the MBR partition table) primary partition on the first
> disk.

But being able to boot only from a primary partition on the first disk
is an unacceptable limitation of the i386 architecture. It just does not
work anymore.

>  When the doz partition is active, it's what control gets passed to,
> where with modern versions its own boot loader can pass control to another
> partition, such as one on which Linux and Grub are installed.

Yeah, I'm doing that, too. It's a simple thing if you know what to do,
but it's not what I would consider "friendly".

But in the end, doesn't it come down to this: all general-purpose
GNU/Linux distros I have seen for many years managed to preserve the
bootability of an preexisting Windows installation, or at least offered
the tools during the install process. Windows just doesn't car and gives
not even an option not to harm the user.