dnf versus yum
vondruch at redhat.com
Mon Jan 6 11:43:50 UTC 2014
Dne 5.1.2014 22:25, Till Maas napsal(a):
> On Sun, Jan 05, 2014 at 01:06:16PM -0600, Chris Adams wrote:
>> Once upon a time, Reindl Harald <h.reindl at thelounge.net> said:
>> Frankly, that's a dumb "feature" to have the package manager know
>> "magic" things about some names. Why is it dumb? Because some people
>> then depend on magic "features". Is this "feature" even documented
>> anywhere? I don't see it in the yum man page for example.
>> This is Unix; system programs are expected to "do what I say". Don't
>> try to code "do what I mean" into it (because what you mean is probably
>> different from what I mean).
>> We've had kernel variant packages in the past, like kernel-smp and
>> kernel-PAE; are all variants supposed to be handled magically? What if
>> there's a new variant? Would not handling it in the package manager
>> magic be a release-blocker bug?
> Kernel packages are special with yum, because multiple packages are
> installed by default. With your argumentation 'dnf update kernel' should
> remove the current kernel when a new kernel is installed. Is this really
> what you expect and what dnf should do? Currently it installs a new
> kernel without removing the old one as I know it from yum.
As far as I know, yum supports installonly packages. While yum supports
them, there is still used special hack for kernel. This hack should be
removed in first place.
Otherwise, I totally agree with Chris and with DNF upstream. "dnf remove
kernel" should remove every kernel and should not behave magically.
More information about the devel