On Thu, May 10, 2007 at 08:57:14PM +0200, Frans Pop wrote: > > > This is only valid for updates, not for new installs. > > So, you would only have to do this _on upgrades_ of grub-pc, not for new > installs. This can be tested separately. Still no need to run anything > for new installs.
We could test if we're upgrading grub-pc, or if we're installing new grub-pc but grub was present before (that one is trickier, any ideas?). It would also be helpful for us to distinguish between the two. Or alternatively, we could set grub-installer to export some variable to indicate its presence. How does that sound? > > We, at least for now, won't install grub2 but make it chainloadable > > using grub and then user can test it and see if he can upgrade or not. > > Sure, but that is only in the short term. In the medium term I think you > should probably offer the choice using a debconf question: install new > grub into MBR or add menu option in current menu.lst to chainload from > current grub. > And IMO you still need debconf questions to explain what is you are doing > as a chainloaded setup is not at all intuitive. I don't think we should offer "install into MBR" for postinst, because this happens before boot with the new core.img has been verified. I'm more inclined for a note that tells the user that we're going to chainload in menu.lst, and that she should install it to MBR using grub-install after verifiing everything is ok. -- Robert Millan My spam trap is [EMAIL PROTECTED] Note: this address is only intended for spam harvesters. Writing to it will get you added to my black list. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
