Big (apparent) reduction in kernel.rpm size

[Patrick O'Callaghan]

On Wed, 2008-07-16 at 00:44 -0400, Mauriat wrote:
> On Wed, Jul 16, 2008 at 12:34 AM, Patrick O'Callaghan <poc at usb.ve> wrote:
> > I just did a "yum upgrade" and happened to notice this (edited highlights):
> >
> >        Installing:
> >         kernel                  x86_64     2.6.25.10-86.fc9  updates            18 M
> >
> >        [...]
> >
> >        Removing:
> >         kernel                  x86_64     2.6.25.6-55.fc9  installed          70 M
> >
> > >From 70M down to 18M? That's a heck of a new optimizer in gcc :-)
> >
> > Just curious.
> 
> Are you comparing installed size (decompressed) vs. download size (compressed)?

I'm just reporting what yum says, but your explanation seems quite
plausible. If yum is using two different criteria for reporting size,
there's a bug in the way it presents information. The "natural"
interpretation of these lines is that installing the new kernel package
will cost me 18M of disk space, while removing the old one will recover
70M, but it looks like what it actually means is that I will recover 70M
and use up an unknown amount greater than 18M.

poc