I've dabbled in making rpm's from tar distributions a bit -- just enough to get a flavor of rpm building. One of the more challenging (frustrating?) rpm tasks is building an rpm for a kernel module. (Or maybe the tar/make package I'm starting from is just not well-designed.) Since the build can be kernel-specific, it's not even clear if building binary rpm's is worth it. Maybe a better idea is to concentrate on making the source rpm easily buildable so that the end user can generate an installation-specific binary rpm and get the benefit of rpm's over tar/make installations. Or, maybe it's not worth rpm's at all for kernel modules. Anyone have opinions on this? TIA, Bill
On Mon, Mar 17, 2003 at 08:23:35AM -0500, Bill Mills-Curran wrote:
I've dabbled in making rpm's from tar distributions a bit -- just enough to get a flavor of rpm building. One of the more challenging (frustrating?) rpm tasks is building an rpm for a kernel module. (Or maybe the tar/make package I'm starting from is just not well-designed.) Since the build can be kernel-specific, it's not even clear if building binary rpm's is worth it. Maybe a better idea is to concentrate on making the source rpm easily buildable so that the end user can generate an installation-specific binary rpm and get the benefit of rpm's over tar/make installations. Or, maybe it's not worth rpm's at all for kernel modules.
Anyone have opinions on this?
A good example of such an RPM is the nvidea drivers. They release a whole slew of different binary ones for different releases of different distros, plus a src rpm that will build to your particular kernel. As long as you have the kernel-source rpm installed that matches the kernel you're running, it works great. http://www.nvidia.com/view.asp?IO=linux_display_ia32_1.0-4191 -- Frank Sweetser fs at wpi.edu WPI Network Engineer
Check out the alsa packages from freshrpms.net. They do exactly what you want. On Mon, Mar 17, 2003 at 08:23:35AM -0500, Bill Mills-Curran wrote: subssn594> I've dabbled in making rpm's from tar distributions a bit -- just subssn594> enough to get a flavor of rpm building. One of the more challenging subssn594> (frustrating?) rpm tasks is building an rpm for a kernel module. (Or subssn594> maybe the tar/make package I'm starting from is just not subssn594> well-designed.) Since the build can be kernel-specific, it's not even subssn594> clear if building binary rpm's is worth it. Maybe a better idea is to subssn594> concentrate on making the source rpm easily buildable so that the end subssn594> user can generate an installation-specific binary rpm and get the subssn594> benefit of rpm's over tar/make installations. Or, maybe it's not subssn594> worth rpm's at all for kernel modules. subssn594> subssn594> Anyone have opinions on this? -- Charles R. Anderson <cra@wpi.edu> / http://angus.ind.wpi.edu/~cra/ PGP Key ID: 49BB5886 Fingerprint: EBA3 A106 7C93 FA07 8E15 3AC2 C367 A0F9 49BB 5886
participants (3)
-
Bill Mills-Curran
-
Charles R. Anderson
-
Frank Sweetser