On Friday, 24 June 2011 03:35:08 David W. Hodgins wrote: > Just today, in alt.os.linux.mandriva, there was a user complaining that > firefox 4 wasn't available. When it was pointed out that it's been > in backports for quite a while, the response was that using backports > "broke" his system.
Where is his bug report? Would QA by the QA team have avoided the problem? (And, the idea with Backports was IMHO that it should not be enabled in the configuration, but used on a case-by-case basis - which was easier once rpmdrake allowed it). > Hopefully Mageia backports won't get that type > of reputation. One way to help with that, is to ensure that all > backports at least have some qa testing. > > I understand your concerns. > > My proposal is to use the backports testing, and see if the qa team > can keep up. If not, having the package pushed directly from cauldron > to backports, or release/updates could be opened up later. Nothing > is being fixed in stone here. In Mandriva, I typically *tested* (minimally) anything I pushed to backports, before I pushed it. But, I certainly don't think having the QA team involved would have provided much value. There was one instance for OpenLDAP where a non-OpenLDAP-maintainer pushed the cooker package to backports without testing, which did create a problem for some users :-(. IMHO, *if* we want to have QA, it should be the onus of users who wanted the backports, and not require any contributor action. E.g. an automatic system based on votes (e.g. require positive feedback from 80% of at least 20 users who have rated a package to promote a package from backports_testing to backports). Regards, Buchan
