On Thu, Apr 27, 2006 at 12:11:39AM +0300, Ville Skyttä wrote: > On Wed, 2006-04-26 at 22:51 +0200, Axel Thimm wrote: > > > The schemes w/o uname-r in the name have been shown to be > > broken some years back. > > The same can be said of the uname-r in package name scheme. No, when I say broken I mean broken as in o removes still required kernel modules o overwrites kernel modules o doesn't allow multiple kernel modules (per kernel) > This is a matter of picking poisons, and depsolvers need special > case adaptations to work sufficiently well with both of the > approaches. The uname-r approach will never remove or overwrite modules. It will keep a kernel rpm happy throughout its lifetime. The only special handling that it requires is installs on kernel upgrades which can be done in a shell one-liner, so depsolver mods should be trivial, too, if needed at all. And w/o the uname-r approach there is simply no way a depsolver can handle it, no matter how much special logic you will try to add to it. You'll end up with further restrictions to accomodate it like "kmdls only for the very latest kernel" or similar. -- Axel.Thimm at ATrpms.net
Attachment:
pgpik7Ktj3ACY.pgp
Description: PGP signature
-- fedora-extras-list mailing list fedora-extras-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-extras-list
