The recent change to packageorder seems to have broken rawhide. This patch makes it work again, but this seems fragile and not compatible with RHEL, nor with any compose that might still have rpms in Product/RPMS/
I also switched to using os.path.join to create the path, seems a little cleaner.
Jesse Keating wrote:
The recent change to packageorder seems to have broken rawhide. This patch makes it work again, but this seems fragile and not compatible with RHEL, nor with any compose that might still have rpms in Product/RPMS/
I also switched to using os.path.join to create the path, seems a little cleaner.
It probably makes more sense to instead to just pass the glob to printMatchingPkg() and have it handle the variations of where packages can be instead of callers (eg, where we do 3 different things for kernel packages)
Jeremy
On Monday 29 January 2007 11:12, Jeremy Katz wrote:
It probably makes more sense to instead to just pass the glob to printMatchingPkg() and have it handle the variations of where packages can be instead of callers (eg, where we do 3 different things for kernel packages)
Worked into a new patch. Seems to work in the Fedora case.
Jesse Keating wrote:
On Monday 29 January 2007 11:12, Jeremy Katz wrote:
It probably makes more sense to instead to just pass the glob to printMatchingPkg() and have it handle the variations of where packages can be instead of callers (eg, where we do 3 different things for kernel packages)
Worked into a new patch. Seems to work in the Fedora case.
Looks good; committed
Jeremy
anaconda-devel@lists.stg.fedoraproject.org
-
Jeremy Katz -
Jesse Keating