On Fri, Apr 17, 2009 at 2:19 PM, Roland McGrath <roland@xxxxxxxxxx> wrote: >> A few of the yum-utils commands take a --archlist, repoquery and >> yumdownloader both do for instance. This is more for looking at data on >> ppc from an i386 box, as i386 is already in the archlist for x86_64. > ^^^ > > You meant x86_64 here, I think. I noticed that switch and noticed that it > didn't help for this case. > >> And that doesn't change what $basearch will be in the config. files >> because doing that is a really bad idea due to the fact that you'll have >> changed what /var/cache/yum/blah/ points to without the next run of yum >> knowing (it's a bad idea to have the same repoid point to more than one >> place). > > Right, I figured that would be the gotcha with what I had in mind. > Maybe it would make sense for yum to name its cache subdirs repo.basearch? > >> So what you probably want is to create a blah-i386 to go with your >> normal blah repo. ... which hardcodes i386 as the arch. Leave it >> disabled, and then whever you want to use it do: > > It seems like a simple enough change to use basearch in the cache dir > names. Then everyone could do this for all the repos without fiddling with > repo configs. Does that seem like something yum could do in the future? > (I'm not asking for any of this to be done next week or anything.) > > Maybe also --repoid et al could grok "reponame.arch" syntax (akin to the > package "name.arch" syntax you can use). Another way to state it is that > the full "repoid" would be "reponame.arch", and "reponame" alone turns into > "reponame.$basearch" implicitly. Not that it helps today, but there is an F12 feature called DebuginfoFS that is supposed to make stuff like this easier. It might be worthwhile to get the mutltilib things sorted out there too. josh -- fedora-devel-list mailing list fedora-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-devel-list
