On Tue, 26 Nov 2002, Randolph Chung wrote: > > Then why doesn't the package itself depend on that kernel? If it did, apt > > would have prevented the upgrade from ever being attempted at all... > > because anyone can compile a kernel themselves and install it, and dpkg > only knows about debian packages.
While anyone indeed can, those who do ought to be shot dead. Reasonable people use always kernel-package to make their own kernels. :) > there might be a better way to check for this by using a preinst check. > Carlos, can you take a look at this? I think sparc already a similar check. This would be a very good idea, indeed. > > I already did, a while back. See the thread when I asked about 712-specific > > hardware and was told not to bother compiling any 2.4.18 kernel anymore. > > 2.4.18? i thought we were talking about 2.4.19.... I had tried 2.4.19, when recommended to do so, then reverted back to 2.4.18 when I noticed how unstable 2.4.19 is on my hardware. My current 2.4.18 resulted in the config I sent to Grant, which he added somewhere to the pa-risc site. > yes, you should not use 2.4.18 anymore. Why? 2.4.18 is the most stable I have seen in ages; it is rock-solid. I've had nothing but troubles with 2.4.19, on but i386 and hppa. Btw, the same libc6 2.3.1 on i386 does not require any particular kernel; it works on 2.4.18 just fine. > The 2.4.19-rc1 kernels are not particularly stable yet, but the pre-rc ones > (e.g. the ones that are in the debian archive is 2.4.19-pa22) should be ok. Noted. I usually build my own kernels, to remove as much excessive stuff as possible, though. -- Martin-�ric Racine "Kas sa tahad mind? - Nej!!! �r du en idiot?!!" http://www.pp.fishpool.fi/~q-funk/ Tallinn, Eesti; Espoo, Suomi.
