> I am still trying to track down what changes in rawhide > [..] caused ghc ABI hashes to mysteriously change. > In the meantime since it is taking longer to resolve than > I had hoped I am requesting an exception from FESCo > to untag the latest broken ghc build from rawhide > since it is not useless anyway unless we do a intermediate rebuild > of all ghc-* packages, but we really need to know > what actually broke rather than sweeping the problem > under a rebuild carpet. A brief status update for those not following [1]: I verified on rawhide that the ABI changes are due to ghc-7.0.1-2.fc15 being built with ghc-6.12.3, whereas ghc-7.0.1-3.fc15 was built with ghc-7.0.1. Ideally this ABI change should not happen but it did this time anyway. (Unfortunately ghc's ABI is not yet 100% deterministic, but upstream wants to continue to improve on ABI stability going forward.) Obviously to allow ongoing updates to the F15 ghc package we have to take the ABI for the build against ghc-7.0.1 as canonical. In future I want to make sure ghc is rebuild against itself after a version update to avoid this problem from recurring. There are more details in the FESCo ticket [1] for those interested but in short we will start rebuilding all the Haskell packages with broken deps now, but it will still be some days before all 85+ packages are rebuilt. Please let me know if you have time to assist with rebuilding on the coming days so we can coordinate efforts. I am also planning to work on automating the rebuild process, to make this less painful in future. Thanks for your patience, Jens [1] https://fedorahosted.org/fesco/ticket/543 -- devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/devel
