On Thu, 2011-03-31 at 07:26 +0200, tim.laurid...@gmail.com wrote: > > Do we need a separate command for this, is it not better to just fix 'yum > update foo-1.2' to do what 'yum update-to foo-1.2', if I understand it > right then 'yum update foo-1.2' will update to the newest version of foo > > 1.2, if foo-1.2 is already installed. This is kind of wrong in my book. > If I do a 'yum install foo-1.2' I expect foo-1.2 to get installed and get a > foo-1.2 is already installed if foo-1.2 is already installed. > yum update foo-1.2 should work the same way.
Well Seth and I discussed it when he saw: http://fossplanet.com/f13/%5Brhelv5-list% 5D-poll-expected-behavior-yum-wheninstalling-updating-packages-108604/ ...and we thought that although although most people expected the "update-to" behaviour changing update directly would almost certainly break something, so we'd do a new command and be safe (we are old gits resistant to change though :). It also helps doing it as a new command if it's broken horribly in some way I didn't test. Doing the one line change to my patch to set update_to=True all the time breaks roughly ~5 test cases ... a lot of them seem to be due to obsoletes processing (AIUI if you have an update and an obsolete "update blah" will install the update, and "update-to blah" will install the obsolete). I am also worried about cross arch. updates (although in quick tests they seem the same), and updates for globs. I can see the desire to "just" fix update, and even if we add the new command now we could "easily" switch over at some future point ... what does anyone else think? _______________________________________________ Yum-devel mailing list Yum-devel@lists.baseurl.org http://lists.baseurl.org/mailman/listinfo/yum-devel
