On Thu, 2005-04-14 at 09:48, Pedro Fernandes Macedo wrote: > And what relation Eclipse has to the lilo issue? None... Pedro: The relevance is that Redhat's excuse for replacing an essential tool (Lilo) with an unreliable/inadequate/undocumented/defunct tool (Grub Legacy) was that Lilo cost too much to support. Redhat is putting a lot of work into Eclipse - a product that few will need or use. A tiny fraction of the Eclipse effort would suffice to retain Lilo in Core. We've shown that Redhat has not supported Lilo in years - Lilo just works. Pretty much all that Redhat needs to do is leave it alone and let it be built and distributed automatically. Redhat then claims that the cost is not in direct Lilo support but rather for massive kludges that are needed in Anaconda and Up2date in order to interface to Lilo. We've shown that there are no such massive kludges - just a generic structure in Grubby that's needed to support half a dozen boot loaders on a variety of different architectures. Redhat makes vague claims that the latest Grub handles RAID-1 and asks us to test Grub yet again. (Perhaps "ask" is an overstatement: Redhat once again pulled Lilo so we'd be forced to test their inferior alternative.) We've shown that Redhat has not provided documentation on using Grub with RAID-1 in any of the logical places, and we've shown that the obvious "grub-install --recheck /dev/md0" does not install Grub to any of the MBR's. Redhat has thus far been unable to tell us any new magic spells for making Grub do what Redhat vaguely implies that Grub can now do. (There are, of course, many complex and error-prone workarounds, but Redhat has been vaguely implying that those are no longer needed.) This has become much more serious than a debate over whether or not someone at Redhat is deliberately trying to cripple their product line. If Redhat does not soon substantiate their claims we will be forced to entertain doubts as to the honesty of the developer of a product which has root privileges on our systems. --Mike Bird