In article <Pine.LNX.4.44.0312290402270.5431-100000@xxxxxxxxxxxxxxxxxxxxx>, Robert P. J. Day <yum@xxxxxxxxxxxxxxxxxxxx> wrote: > # yum list "*ogle*" > Gathering header information file(s) from server(s) > Server: Fedora 1 - Base > Server: Fedora 1 - Stable > Server: Fedora 1 - Testing > Server: Fedora 1 - Unstable > Server: Fedora 1 - Updates > Server: Fedora 1 - Updates testing > Server: Freshrpms 1 - Freshrpms > Server: Freshrpms 1 - Testing Updates > Server: Freshrpms 1 - Updates > Server: Livna 1 - updates > Finding updated packages > Downloading needed headers > Looking in Available Packages: > Name Arch Version Repo > ----------------------------------------------------------------------- > ogle i386 0.9.2-1.1.fr freshrpms-freshrpms > ogle i686 0.9.2-0.lvn.2.1 livna-updates > ogle-devel i386 0.9.2-1.1.fr freshrpms-freshrpms > ogle-devel i686 0.9.2-0.lvn.2.1 livna-updates > ogle_gui i386 0.9.2-1.1.fr freshrpms-freshrpms > > Looking in Installed Packages: > Name Arch Version Repo > ------------------------------------------------------------------ > > so while both freshrpms and livna claim to supply both ogle and > ogle-devel, since i use a pkgpolicy of "newest", one would assume that i'd > get the freshrpms versions in both cases. (i'm assuming that the > comparison takes the build value into account, yes?) Is it possible that yum was preferring arch==i686 over arch==i386 at a higher precedence than the version number? Cheers, Tony -- Tony Mountifield Work: tony@xxxxxxxxxxxxx - http://www.softins.co.uk Play: tony@xxxxxxxxxxxxxxx - http://tony.mountifield.org