My only desire is to get the exact source that built the kernel (w/o manual labor or version-interpolation) - an rpm is fine, so is tar, etc. But once I have it I have no motivation for repackaging it. It will likely live on just one machine.On Fri, Oct 15, 2004 at 02:45:02PM -0600, Ken Johanson wrote:
I meant to say that its desirable to get the incremental updates using any lightweight (or not) methods, say CVS, without downloading a whole new rpm (source tree) for each update - or even becoming dependent on rpm...
This is a branch into a different argument -- if you don't want to use RPM, why do you want a binary RPM of the kernel source?
But really, I think you _do_ want to use RPM, and just don't know it yet. :)
Opinions vary on how to maintain/distro kernels across multiple machines, but given that most machines dont have the same hardware (even in large enterprise server-rooms) or even roles (need for a given module), and given that its a just good feeling to have a build-able source tree living on every machine (space is not an issue for me lately), I cant make much use of rpm for kernel management - I agree theres some great capabilities in it though for managing versions in applications though. Maybe I'll find the kernel beauty in it too, someday. Right now I'm more worried about its inital availability, rpm or no. :-)
