On Mon, 11 Mar 2013 03:49:38 +0100 Kevin Kofler <kevin.kofler@xxxxxxxxx> wrote: > Miloslav Trmač wrote: > > If a new package doesn't break tests, it will tagged into rawhide > > immediately or overnight - just like now. No extra work needed, no > > change in workflow. > > Running the tests alone will slow down chain builds significantly, > even if the builds get tagged immediately after the tests pass. Well, not sure it would. If builds were tagged into f19-pending and the tests ran from that, then tagged into f19, the max delay would be when a newrepo just started and it has to wait for the next newrepo to be added. The min delay would be that it gets added and newrepo starts right then. So with current newrepo tasks about 13min per newrepo, so thats 13 or 26 min per build. > > I don't think there is any obvious reason to exempt anaconda from > > this process. Of course, we start with zero tests, and thus also > > zero requirements on anaconda; OTOH I expect somebody will propose > > to add a test that "minimal network install should always be > > working" very soon after the infrastructure is ready. Such a test > > would obviously apply both to anaconda updates and to changes of > > everything else that may break anaconda. > > It's quite funny that you're talking about enforcing requirements on > Anaconda AFTER FESCo gave Anaconda developers carte blanche to put a > KNOWN BROKEN Anaconda into pre-F18 Rawhide and leave it there even > when it was known not to get fixed in time for the release, causing > release slippage. Does this mean you admit that FESCo made a mistake? > Or that FESCo as a whole finally admits to having made a mistake? :-) <eye-roll> Hasn't this been discussed to death? We could have taken the hit in f18 and landed the new anaconda, or we could have reverted and taken the hit in f19 to land the new anaconda, etc. It had to be done, sooner is better than latter. kevin
Attachment:
signature.asc
Description: PGP signature
-- devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/devel
