[Yum-devel] Even fancier yum 3.2.11 drain bammage
Hans-Peter Jansen
hpj at urpla.net
Wed Feb 13 23:28:39 UTC 2008
Am Mittwoch, 13. Februar 2008 schrieb seth vidal:
> On Wed, 2008-02-13 at 23:19 +0100, Hans-Peter Jansen wrote:
> > Am Mittwoch, 13. Februar 2008 schrieb seth vidal:
> > > On Wed, 2008-02-13 at 22:06 +0100, Hans-Peter Jansen wrote:
> > > > # yum update
> > > >
> > > > Dependencies Resolved
> > > >
> > > > ===================================================================
> > > >==== ====== Package Arch Version
> > > > Repository Size
> > > > ===================================================================
> > > >==== ====== Installing:
> > > > kernel-pae i586 2.6.24.1-35.1 kernel-head
> > > > 22 M replacing kernel-bigsmp.i586 2.6.18.8-0.5
> > > >
> > > > kernel-source i586 2.6.24.1-35.1 kernel-head
> > > > 50 M kernel-syms i586 2.6.24.1-35.1
> > > > kernel-head 1.3 M Installing for dependencies:
> > > > kernel-bigsmp i586 2.6.18.2-34 base
> > > > 18 M
> > > >
> > > > Transaction Summary
> > > > ===================================================================
> > > >==== ====== Install 4 Package(s)
> > > > Update 0 Package(s)
> > > > Remove 0 Package(s)
> > > >
> > > > Total download size: 92 M
> > > > Is this ok [y/N]:
> > > >
> > > > It's not, off cause.
> > > >
> > > > Even if kernel-pae does the usual provides/obsoletes kernel-bigsmp,
> > > > shouldn't that be ignored in the installonlypkgs case? It doesn't
> > > > make much sense in my book. Needless to say, the replacement is
> > > > arbitrary:
> > >
> > > it looks like kernel-pae is obsoleting kernel-bigsmp but you still
> > > have a dangling requirement that kernel-pae isn't providing. So yum
> > > is, correctly, pulling in the only thing available for it.
> >
> > Well, if that's the case, then yum should replace all kernel-bigsmp
> > packages, not just an arbitrary one, but I cannot find any reason, why
> > it should do this at all, if the package is in _installonly_ mode. At
> > least, this is how I understand such a mode.
> >
> > > This looks like kernel-pae is not adequately providing what it is
> > > obsoleting.
> >
> > Hmm, all it obsoletes, it also provides:
> > Provides: k_smp4G kernel-bigsmp
> > Obsoletes: k_smp4G kernel-bigsmp
>
> Can you give me a dump of the requires from the nvidia pkg which drug
> kernel-bigsmp back in?
Sure:
> rpm -qR nvidia-kmp-bigsmp-100.14.11_2.6.18.8_0.5-1
rpmlib(VersionedDependencies) <= 3.0.3-1
kernel-bigsmp
/bin/sh
/bin/sh
/bin/sh
rpmlib(PayloadFilesHavePrefix) <= 4.0-1
rpmlib(CompressedFileNames) <= 3.0.4-1
kernel(vmlinux) = bb95ba9d667bafff
kernel(drivers_char_agp) = 03f4fd294a99e70a
kernel(drivers_i2c) = 5e5b84e3abb011dc
rpmlib(PayloadIsBzip2) <= 3.0.5-1
Oh well, I see and remember, that the whole kernel module dependency
processing is handled by some obscure macros, I already filed a bug report
in the Novell Bugzilla, because of some obvious bugs (without response).
Here. it's the kernel-bigsmp line without specifying the version, I can
imagine.
Thanks,
Pete
More information about the Yum-devel
mailing list