On Tue, May 02, 2006 at 06:22:49PM -0600, Archaic wrote: > On Wed, May 03, 2006 at 01:07:49AM +0100, Ken Moffat wrote: > > > > At the moment, we seem happy to let people stick with the kernel they > > built when they built LFS. > > I don't understand. We don't control what a user does. We give a build > recipe, throw in some educational material, give guidelines on how to > maintain the system. What exactly are you asking that we do? > I guess I'm suggesting we should encourage people to upgrade! In particular, people continue to build the released version of the book months, or years, afterwards.
> 2.6.16 is most likely the kernel that will be used in lfs-6.2 due mostly > to timing issues. We can't wait for 2.6.17 to settle down, else a book > would never get released. If a 2.6.16.x release comes out during the > testing phase, we could look at it, test it a bit, perhaps discuss > on-list, but I'm not prepared to make the blanket statement that any > point release from the 2.6.16 branch is a suitable drop-in without > testing. My argument for -stable kernels is that they are intrinsically better (because they fix identified bugs), and that until the book is released we should pick up any new versions [ for 2.6.16 in this case ] without waiting for an editor to build with them. Some of the bugs are probably trivial - most LFS users won't have been exposed to the nfs data corruption in early 2.6.16 [ fixed in 2.6.16.3 or earlier ], and compile failures on other architectures should not concern us here, but any suggestion that the testing resources available to us prove a particular kernel version is "stable" seems optimistic to me. But I agree totally that we shouldn't wait for 2.6.17 - it has been long enough since the last release, and anyway by then we can look forward to what will be in 2.6.18. Ken -- das eine Mal als Tragödie, das andere Mal als Farce -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
