yum annoyances

Jeff Spaleta jspaleta at gmail.com
Tue Dec 6 20:42:07 UTC 2005


On 12/6/05, Matthew Miller <mattdm at mattdm.org> wrote:
> I agree with you that this shouldn't happen in updates to released versions
> of Fedora.



> If it's known that package is going to be added to extras at some point
> soon, that package probably shouldn't be added to the obsoleted-packages
> list.

As soon as we get that read-the-future plugin into the Extras
buildsystem. As an exercise for the reader... how many dropped
packages in FC4 that were in FC3 were submitted for review before FC4
release? How many were submitted after FC4 release?


> I'm more concerned about the ones that do have broken deps, which seems to
> be a lot of the time in cases like this.

If installed packages have broken deps.... hasn't anaconda in the past
advised you of this as part of the install? fc5t1 anaconda doesn't
count since its clearly not representative of full anaconda
functionality. I don't see why this is a concern unless anaconda is
going to start handling upgrade situation differently.

> It still could be; the obsoletes-packages RPM would just be a tool to help
> if you decide to go that way.

If you install obsoletes-packages... there is no decision point. If
obsoleteing is active in the transaction,  it obsoletes all the
packages that is set to obsolete.. like all other obsoletes in all
other packages. For this package to be a tool to help you "decide", it
would have to be treated differently than all other obsoletes in all
other packages which obsolete without user interaction. Are you know
suggesting that this mythical obsoletes-package be treated specially
by the tools?

-jef




More information about the fedora-devel-list mailing list