source rpm: granule-1.2.4-2.fc7.src.rpm package: granule - 1.2.4-2.fc7.i386 from fedora-8-i386 unresolved deps: libassa-3.4.so.0 Why has libassa not been pushed to testing first?
My bad. I misread the instructsions. The newly released libassa-3.5.0 should have been deployed not as an update to libassa-3.4.0, but rather as a separate package since they can be installed in parallel. My intention was to upgrade granule-1.2.4 to granule-1.3.0 which requires libassa-3.5.0. Should I just remove granule-1.2.4 from fc8? best, --Vlad
I don't understand your plans. Why remove anything if you want to upgrade it anyway? In a case like this, simply upgrade both libassa and granule _at once_. You can tell bodhi, the Fedora Updates System, to push a _set_ of packages. [...] In general, if a library is API- or ABI-incompatible (as in the case of the new libassa) and has not been released into the Fedora repositories yet, mail releng and request that the build in koji is tagged into the buildroot and made available for subsequent build jobs. Then you can build any dependencies and push all results at once in a good set that doesn't break. Since the new libassa is in the updates repository already, you don't need to mail releng this time.
I wasn't aware that you can push a _set_ of packages. Thanks for the tip. "... mail releng and request that the build in koji is tagged into the buildroot and made available for subsequent build jobs." I guess I missed 'releng' feature alltogether - it must have been described somewhere. thanks for your advise.
I can confirm that updating granule from the updates-testing repo works as expected. Is it ready to be pushed to the updates repo? $ sudo yum --enablerepo=updates-testing update granule ... Updated: granule.i386 0:1.3.0-0.fc8 libassa.i386 0:3.5.0-1.fc8 Complete!
Yes, it is. I have done a fair amount of testing myself both on FC-7 and FC-8 systems and everything seems to be in place and working.
Is this still a bug? If not, can we close it?
Yes, the granule update does not show up in the broken deps. But that is something you can verify easily yourself.