It was thought that rebuilding ghc-vector would fix the hash issues on aarch64. It appears not. DEBUG util.py:378: Getting requirements for gtk2hs-buildtools-0.13.0.4-1.fc23.src DEBUG util.py:378: --> alex-3.1.4-2.fc23.aarch64 DEBUG util.py:378: --> ghc-Cabal-devel-1.18.1.5-46.fc23.aarch64 DEBUG util.py:378: --> ghc-array-devel-0.5.0.0-46.fc23.aarch64 DEBUG util.py:378: --> ghc-containers-devel-0.5.5.1-46.fc23.aarch64 DEBUG util.py:378: --> ghc-directory-devel-1.2.1.0-46.fc23.aarch64 DEBUG util.py:378: --> ghc-filepath-devel-1.3.0.2-46.fc23.aarch64 DEBUG util.py:378: --> ghc-hashtables-devel-1.1.2.1-2.fc23.aarch64 DEBUG util.py:378: --> ghc-pretty-devel-1.1.1.1-46.fc23.aarch64 DEBUG util.py:378: --> ghc-process-devel-1.2.0.0-46.fc23.aarch64 DEBUG util.py:378: --> ghc-random-devel-1.0.1.1-30.fc23.aarch64 DEBUG util.py:378: --> ghc-rpm-macros-1.4.15-3.fc23.aarch64 DEBUG util.py:378: --> happy-1.19.5-3.fc23.aarch64 DEBUG util.py:378: Error: Package: ghc-hashtables-1.1.2.1-2.fc23.aarch64 (build) DEBUG util.py:378: Requires: ghc(vector-0.10.9.1-141b8608a457666e762b17e65827314c) DEBUG util.py:378: Error: Package: ghc-hashtables-devel-1.1.2.1-2.fc23.aarch64 (build) DEBUG util.py:378: Requires: ghc-devel(vector-0.10.9.1-141b8608a457666e762b17e65827314c) DEBUG util.py:378: You could try using --skip-broken to work around the problem DEBUG util.py:378: You could try running: rpm -Va --nofiles --nodigest
I guess one question is: is it still possible to reproduce the above old ABI hash (ghc-vector-0.10.9.1-1.fc22) sometimes when building ghc-vector for aarch64? Both ghc-vector-0.10.9.1-2.fc23 and ghc-vector-0.10.9.1-3.fc23 provide ghc(vector-0.10.9.1-34f56500c95b94cf9d44532097ee0c58). Vector is used by about 30 packages in Fedora.
Some scratch builds to try to see ABI instability: http://arm.koji.fedoraproject.org/koji/taskinfo?taskID=3141234 http://arm.koji.fedoraproject.org/koji/taskinfo?taskID=3141242 http://arm.koji.fedoraproject.org/koji/taskinfo?taskID=3141247
(In reply to Jens Petersen from comment #2) > http://arm.koji.fedoraproject.org/koji/taskinfo?taskID=3141234 > http://arm.koji.fedoraproject.org/koji/taskinfo?taskID=3141242 > http://arm.koji.fedoraproject.org/koji/taskinfo?taskID=3141247 All 3 have 34f56500c95b94cf9d44532097ee0c58. http://arm.koji.fedoraproject.org/koji/taskinfo?taskID=3143067 http://arm.koji.fedoraproject.org/koji/taskinfo?taskID=3143069
The new hash seems stable so far. http://arm.koji.fedoraproject.org/koji/taskinfo?taskID=3143436 The main thing changed between the builds is ghc I guess: * Thu Jun 11 2015 Jens Petersen <petersen> - 7.8.4-45 - use ld.gold on aarch64 like for armv7 (Erik de Castro Lopo, #1195231) * Wed Apr 22 2015 Jens Petersen <petersen> - 7.8.4-44 - turn on SMP and ghci for aarch64 (Erik de Castro Lopo, #1203951) Though in principle neither should affect the ABI think? Dunno if it is some other part of the toolchain. Anyway currently I see no choice but to rebuild the 30 packages above that consume ghc-vector... ;-( (Given that they are uninstallable on aarch64 I don't see the problem with rebuilding the NVRs just for aarch64 but i guess we have to follow the rules: wouldn't want to make life easy on ourselves.;) But maybe it is easier anyway to let shadowbuild handle it all anyway. :)
> Anyway currently I see no choice but to rebuild the 30 packages above that > consume ghc-vector... ;-( Which 30? What NVRs, if the the latest NVRs have already been built on aarch64 we need them bumped and built on mainline and submitted as updates.
Do you have a list of packages that are currently failing to build or install? (The number 30 I just got trying to run "rpm -e ghc-vector" on my f22 laptop.)
(In reply to Jens Petersen from comment #6) > Do you have a list of packages that are currently failing to build or > install? > > (The number 30 I just got trying to run "rpm -e ghc-vector" on my f22 > laptop.) So dependencies. See the issue above, do we need to rebuild ghc-hashtables then in that example?
ghc-hashtables-1.1.2.1-3.fc23 ghc-aeson-0.8.0.2-3.fc23 ghc-yaml-0.8.10-3.fc23 ghc-dbus-0.10.10-3.fc23 ghc-pandoc-types-1.12.4.1-3.fc23 ghc-JuicyPixels-3.2.2-3.fc23 has been submitted as an update to Fedora 23. https://bodhi.fedoraproject.org/updates/FEDORA-2015-14721
ghc-JuicyPixels-3.2.2-3.fc23, ghc-aeson-0.8.0.2-3.fc23, ghc-dbus-0.10.10-3.fc23, ghc-hashtables-1.1.2.1-3.fc23, ghc-pandoc-types-1.12.4.1-3.fc23, ghc-yaml-0.8.10-3.fc23 has been pushed to the Fedora 23 testing repository. If problems still persist, please make note of it in this bug report.\nIf you want to test the update, you can install it with \n su -c 'yum --enablerepo=updates-testing update ghc-JuicyPixels ghc-pandoc-types ghc-aeson ghc-hashtables ghc-dbus ghc-yaml'. You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2015-14721
pandoc-1.13.2-4.fc23 ghc-texmath-0.8.0.1-4.fc23 has been submitted as an update to Fedora 23. https://bodhi.fedoraproject.org/updates/FEDORA-2015-14762
ghc-texmath-0.8.0.1-4.fc23, pandoc-1.13.2-4.fc23 has been pushed to the Fedora 23 testing repository. If problems still persist, please make note of it in this bug report.\nIf you want to test the update, you can install it with \n su -c 'yum --enablerepo=updates-testing update pandoc ghc-texmath'. You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2015-14762
ghc-texmath-0.8.0.1-4.fc23, pandoc-1.13.2-4.fc23 has been pushed to the Fedora 23 stable repository. If problems still persist, please make note of it in this bug report.
ghc-JuicyPixels-3.2.2-3.fc23, ghc-aeson-0.8.0.2-3.fc23, ghc-dbus-0.10.10-3.fc23, ghc-hashtables-1.1.2.1-3.fc23, ghc-pandoc-types-1.12.4.1-3.fc23, ghc-yaml-0.8.10-3.fc23 has been pushed to the Fedora 23 stable repository. If problems still persist, please make note of it in this bug report.