On Sat, Nov 17, 2007 at 04:52:12PM -0600, Bruce Dubbs wrote: > Ken Moffat wrote: > > > I suggest that the command in the book should be '--enable-current' > > I thingk the option you are referring to is --enable-kernel=current, not > --enable-curent. > Well spotted, Bruce. Thanks. > My suggestion is to use, --enable-kernel=2.6.11.12. The reason I choose > that version is that it is the version in LFS 6.1.1. That version of > LFS is the oldest in SVN and that version of the kernel was released > 12-Jun-2005. > > This should be old enough for almost any halfway recent distro. I do > not see an overriding benefit to current. It just seems like an > aggressive optimization to me. > Old enough, yes. Appropriate to getting the "goodies" from the recent versions of glibc, no. Anyway, the key word in what you have written is "almost" - we can't make guarantees about what somebody is trying to build from, except to say that it ought to meet our minimum requirements.
Also, correct me if I'm wrong, but version magic numbers are in the x.y.z form as far as glibc is concernd, I don't believe they get incremented for -stable releases. > I do agree that the books should discuss this option in a bit more > detail than it does now to explain the pros and cons and to mention the > =current option. If we pick a variant of the "oldest practical" kernel, why do we bother to upgrade glibc (other than to get timezone changes and bug fixes) ? In my experience, LFS has always tried to use recent versions of software, and it now looks as if building for recent versions of the kernel is a prerequisite to making the most of glibc (indeed, has been preferred for a long time, it's just that we didn't realise). ĸen -- 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
