On Fri, 24 Oct 2008 09:23:10 -0700, Darren New wrote:

> Jim Henderson wrote:
>> I was hoping I hadn't missed something with GRUB development. ;-)
> 
> In matters Linux, always assume I don't know more than you do. :-)

Thing is, I always assume I'm going to learn something new, especially 
when talking with you, Darren. :-)

>>>>> Maybe. I don't have any trouble using Windows' bootloader to load
>>>>> GRUB. You just have to set it up right. It's pretty trivial.
>>>> Yup.  But it's not something that I've ever seen automated by the
>>>> Linux installers.
>>> No, because they can't write to NTFS partitions.
>> 
>> ntfs-3g can write safely to NTFS partitions.
> 
> Yes, but since (as I understand it) you need a copy of ntfs.sys on your
> Linux partition to make that work, Linux first has to already be booted
> and capable of mounting the right drive and finding ntfs.sys to copy it.
> I guess you could do it, but it certainly sounds ... messy. :-)

I'm now wondering if that's the case - if it uses ntfs.sys; I was under 
the impression it does, but in searching for that information, it seems 
another driver, called the "captive" driver uses a shim from ReactOS to 
wrap around ntfs.sys.  I'll have to do a little more digging to see if 
ntfs-3g uses that or not.

>> Yeah, it's really kinda - well, expected, I guess - that MS behaves
>> that way about how the system boots.
> 
> Dude, it's been that was since the IBM XT came out.  The very first hard
> drive, before you could have more than 4 partitions, had a boot sector
> that would boot off the "active partition". That's what it's *for*.

Well, yes.  There was specific boot sector code in the MBR to do that, 
though.  GRUB can be installed either way, but it does enhance things a 
bit more - kinda like the old System Commander software did.

> Yes, MS's boot sector follows the standard that's been around for a
> decade longer than Linux has. GRUB's doesn't. What do you think MS is
> doing wrong here?

My understanding is that the Microsoft MBR (at least as included in 
Vista, possibly with older versions as well) depends on that boot.ini 
file, much as GRUB depends on the files in /boot/grub.

>> They want to be the only OS there, so they just play like they are. 
>> They should stop doing that, maybe they will with Windows 7?
> 
> Doing what? I have no problem at all telling Windows to boot Linux on
> the next go, simply by changing the active partition to be partition 2
> instead of partition 1.  Once I get into Linux, I can reconfigure GRUB
> to default to booting Windows. But after that, I can never get back into
> Linux again, because Windows can't change GRUB's configuration, so if I
> set the active partition to the GRUB partition, it'll still boot
> Windows.

Thing is, it shouldn't require a change to the active partition list.  
But also, GRUB can be told to change the default for a single boot only - 
at least I seem to recall there's an option to do that.

> Now who is the only OS that'll boot?

Installing Linux (at least openSUSE) on a drive with Windows on it, the 
installer will set Windows up as a menu option so you can select either.  
Installing Windows after Linux, though, Windows won't add Linux to the 
boot menu automatically.  I guess that's what I was trying to say.

> If GRUB's boot sector actually loaded the boot sector off the active
> partition, just like ever other boot sector written in the history of
> hard drives on IBM computers and their clones, you could remotely switch
> back and forth by using the same procedure that's worked since day 0 on
> the IBM AT. Sadly, the Linux developers want to be the only OS that can
> be booted, so they didn't bother to follow the standard. ;-)

I had thought GRUB on my system here was set up with the MBR, but I was 
in fact mistaken - it's in the root partition instead.

Jim

Post a reply to this message