On Mon, 22 Sep 2003, Hedemark, Magnus wrote: > rpjday said: > > > **** Installation > > - list of pre-requisite RPMs here > > python > rpm-python > rpm >= v4.1.1 > libxml2-python sorry, i just meant that i was *going* to list the pre-reqs there. just being lazy. > > For advanced yummers (and covered later somewhere), you can > > build your own repositories with > > > > /usr/bin/yum-arch > > Don't forget to mention yum-arch is expecting an argument. i wouldn't be expecting to even get *remotely* into yum-arch at this point, even to the extent of saying it needed an argument. i'd save that for later. > > (is it worth suggesting turning off the yum daemon for the > > time being until one gets more familiar with running it > > manually?) > > I don't see the point of it with the cron job that is installed in > /etc/cron.daily/yum.cron (or am I missing something?) but turning off yum removes the lock file so that the cron job doesn't actually end up doing anything -- that's what i was getting at. or am i misunderstanding something? # service yum status # service yum stop # chkconfig yum off that sort of thing. i just wanted to emphasize that one shouldn't let yum operate automatically until one is comfortable with how it works manually. > > (question: AFAICT, a repository must have a top-level > > headers/ directory, but the associated RPMs directory > > need not be at the same level. what's the flexibility here?) > > Seems to be fairly flexible. The way I have it set up now, there is a > repository root with a headers/ directory in it, and below the repository > root there may be a number of nested subdirectories with packages. yum-arch > will index the relative paths and also what architecture each package is > built for so the client can choose the package that best fits. hmmm ... i'm going to have to look at that a bit more carefully. i suspect figuring out how yum-arch works would help. all i wanted was to explain to a newbie what a repository directory structure might look like. rday