On Fri, Jul 02, 2004 at 04:31:52PM -0400, Derek Atkins wrote: > > Also along those lines -- do you see an advantage of having a separate > > openafs-kernel subpackage? Why not just include that in -client? > Yes, there's an advantage. Indeed, my plan for the "official"
Let me be more specific. :) What advantage do you see in having a separate package? In most cases, they won't be built or upgraded separately from the client package anyway, will they? > openafs-1.4 RPMS is to have separate openafs-kernel RPMS for each > kernel version. Hmmm. That seems like a bit more overhead -- each new kernel version will require changes at several places in the spec file. Each a cut & paste, but with various lines changed.... But more importantly, how will that work for upgrades? I don't think you can have multiple subpackages with the same name but different versions, so the kernel version would have to be encoded in the subpackage name itself. Each one could provide "openafs-kernel-module" or some other virtual name, and each could require its corresponding kernel package, but how would the packaging system know that a new version of the afs kernel module should be installed when a new kernel is? -- Matthew Miller [EMAIL PROTECTED] <http://www.mattdm.org/> Boston University Linux ------> <http://linux.bu.edu/> _______________________________________________ OpenAFS-devel mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-devel
