> can you give me an example? I never see (or just not recognize it?) > I usualy do kernel and glibc by hand and all other by rpm -Fvh * multi archs was one :) iirc it didn't handle them well, maybe it does now... > it's not a solution in a production enviroment! eg in case of kernel you > should have to reboot ASAP since some kernel modul can't loaded if it > just required after the update! another example our samba server > _should_ have to run 24 hour a day it can be stopped on saturday between > 20-24. ok I know about exclude but there is a few server on which I > would like to select the packages manualy (since I belive in myself > which is the biggest mistake) but after I select them I'd like to update > them in a branch. not really true. yum installs kernels, always. It will never update a kernel. So you'd never be in that state with yum. ie: the modules for old kernels are still there. It'd be unsafe to do otherwise. you should look into some of the options that people have been coming up with. what if you had yum use a url for its config file? yum -c http://myserver/yum.cgi then used the command= option in the conf file to tell it what to do? then you could time certain updates for friday or saturday and keep your system safe - completely w/o entering an 'excludes' in. One thing that is higher on my list of things would be to be able to match package updates with "Actions" so if yum notices that pkg x changed state it would take an action based on the change of state. the action could be a user written script sorta like something BEYOND what %post and %pre inside an rpm do. > this is just a matter of taste, but I'm still lazy:-((( join the club - laziness prompted me to write yum ;) -sv
