Understanding Prioritization

Gustavo Niemeyer gustavo at niemeyer.net
Wed Dec 21 12:22:19 PST 2005


Hi Eli,

[...]
> I really hope smart will handle this the way I'm expecting. I really
> want to get past the impending disappearance of apt-get for RPMs. :(

Before trying to make it work exactly the way you'd expect it,
I'd really appreciate if you could try the way it was designed
to work.

The example below is a nice test.

> APT-GET REPOS AND PRIORITIES
> kde-redhat repos 999
> freshrpms           998
> fc4's                   default
> 
> APT-GET UPDATE UPGRADE PRODUCES
> The following packages will be upgraded
>    fetchmail (6.2.5.4-1.fc4 => 6.2.5.5-1.fc4)
>    kdegraphics (3.5.0-1.4.fc4.kde => 3.5.0-2.1.fc4.kde)
>    kdegraphics-devel (3.5.0-1.4.fc4.kde => 3.5.0-2.1.fc4.kde)
>    libmusicbrainz (2.1.2-0.1.fc4.kde => 2.1.2-0.2.fc4.kde)
>    libmusicbrainz-devel (2.1.2-0.1.fc4.kde => 2.1.2-0.2.fc4.kde)
>    mysql (4.1.15-1.FC4.1 => 4.1.16-1.FC4.1)
>    mysql-bench (4.1.15-1.FC4.1 => 4.1.16-1.FC4.1)
>    mysql-devel (4.1.15-1.FC4.1 => 4.1.16-1.FC4.1)
>    mysql-server (4.1.15-1.FC4.1 => 4.1.16-1.FC4.1)
>    postgresql (8.0.4-2.FC4.1 => 8.0.5-1.FC4.1)
>    postgresql-contrib (8.0.4-2.FC4.1 => 8.0.5-1.FC4.1)
>    postgresql-devel (8.0.4-2.FC4.1 => 8.0.5-1.FC4.1)
>    postgresql-docs (8.0.4-2.FC4.1 => 8.0.5-1.FC4.1)
>    postgresql-libs (8.0.4-2.FC4.1 => 8.0.5-1.FC4.1)
>    postgresql-pl (8.0.4-2.FC4.1 => 8.0.5-1.FC4.1)
>    postgresql-python (8.0.4-2.FC4.1 => 8.0.5-1.FC4.1)
>    postgresql-server (8.0.4-2.FC4.1 => 8.0.5-1.FC4.1)
>    postgresql-tcl (8.0.4-2.FC4.1 => 8.0.5-1.FC4.1)
>    postgresql-test (8.0.4-2.FC4.1 => 8.0.5-1.FC4.1)
>    xmms-mp3 (1.2.10-0.lvn.3.4 => 1.2.10-13.2.fc4
> 
> 
> SMART REPOS AND PRIORITIES
> kde-redhat  0
> freshrpms  -1
> fc4's          -5
> 
> SMART UPDATE UPGRADE PRODUCES
> Upgrading packages (4):
>   kdegraphics-7:3.5.0-2.1.fc4.kde at i386
>   kdegraphics-devel-7:3.5.0-2.1.fc4.kde at i386
>   libmusicbrainz-2.1.2-0.2.fc4.kde at i386
>   libmusicbrainz-devel-2.1.2-0.2.fc4.kde at i386

Notice that the APT transaction above is much more expensive than
the transaction below. As I told you, Smart considers the
priority *difference* in certain occasions. Can you please try
to increase the difference of these repositories to reflect what
you want to happen? 1 and 5 priority points is not really a
difference big enough to change the transaction above the way you
expect.

Please, let me know how it goes.

-- 
Gustavo Niemeyer
http://niemeyer.net



More information about the Smart mailing list