weird dependency between openoffice.org from fedora and oxygen-openoffice in F9

Michael Schwendt mschwendt at gmail.com
Fri Aug 15 15:18:47 UTC 2008


On Fri, 15 Aug 2008 16:51:37 +0200, joachim.backes at rhrk.uni-kl.de wrote:

> Michael Schwendt wrote:
> > On Fri, 15 Aug 2008 13:50:04 +0200, joachim.backes at rhrk.uni-kl.de wrote:
> > 
> >> I installed Oxygen-openoffice (not from the fedora repos) in F9.
> >>
> >> Later on, "yum update" tried to update openoffice.org from the fedora repos!
> >>
> >> I cancelled this, so that my oxygen software remained clean!
> >>
> >> Question: Why did yum try to update the oxygen packages?
> >>
> >> Installed rpms:
> >> ===============
> >>
> >>   rpm -qa '*openoffice.org*'
> >> openoffice.org-core08-2.4.1-9310.i586
> >> openoffice.org-emailmerge-2.4.1-9310.i586
> > 
> > Can you do this query again but *including* the Epoch values?
> > $ rpm -qa --qf %{e}:%{n}-%{v}-%{r}\\n '*openoffice.org*'
> 
> Hi Michael, here they are:
> ==========================
> 
>   rpm -qa --qf %{e}:%{n}-%{v}-%{r}\\n '*openoffice.org*'
> (none):openoffice.org-core08-2.4.1-9310
> (none):openoffice.org-emailmerge-2.4.1-9310
  ^^^^^^
All have no Epoch, but the Fedora rpms of openoffice.org have Epoch 1 (you
can also see that in Yum's output -- it's the "1:" at the beginning).
Therefore they are considered as the "newer" packages. In RPM version
comparison, the hidden Epoch value is most-significant. Epoch comparison
overrides the result of ordinary version-release comparison. Highest
Epoch wins, and any non-zero Epoch wins over a missing Epoch.




More information about the users mailing list