Package: scotch Followup-For: Bug #1121516 X-Debbugs-Cc: [email protected] User: [email protected] Usertags: ppc64el Control: fixed -1 7.0.10-5 Control: unblock 1121012 by -1 Control: severity -1 normal
The ppc64el problem in scotch has been fixed in 7.0.10-5 after identifying that -DCOMMON_PTHREAD needs to be provided in order to allow the tests to pass. (It's strange to me then that only ppc64el failed when this flag was not set. I would have thought other architectures would be affected by the same problem) The upgraded library chain from scotch to petsc/slepc has now migrated to testing, so this bug is no longer affecting the scotch-petsc transition bug#1121012. Hence removing the block. I'm leaving this bug open for now pending advice from upstream on the best way to manage these pthread flags. Part of the issue to discuss is whether upstream likes the idea of running standalone CI tests. Upstream set up build-time tests only, it was a debian patch enabling separate debci tests that triggered this bug. So the question is whether -DCOMMON_PTHREAD should be provided manually as we're now doing in debian/tests, or whether it should be added automatically in cmake scripts for standalone tests (just as it's added in the toplevel cmake build scripts). Also we'll want to remove the ppc64el debug code from debian/rules and debian/tests before closing this bug.
