Public bug reported: pandas 2.3.3+dfsg-3ubuntu2 fails to build in stonking-proposed on amd64, amd64v3, arm64 and ppc64el. Retrying gives the same result.
One test fails, and it fails the same way on all four architectures: FAILED pandas/tests/io/pytables/test_store.py::test_no_track_times AssertionError: assert checksum_0_tt_false == checksum_1_tt_false = 1 failed, 12343 passed, 2965 skipped, 97 deselected, 155 xfailed, 1 xpassed = https://launchpadlibrarian.net/880061555/buildlog_ubuntu-stonking- amd64.pandas_2.3.3+dfsg-3ubuntu2_BUILDING.txt.gz Cause: The test writes an HDF5 file twice with track_times=False and expects the two files to be byte-identical. With hdf5 2.2.0+repack-5, which has been in stonking since August, the output is no longer deterministic. The previous successful pandas build (ubuntu1, April) used hdf5 1.14. armhf, riscv64 and s390x build fine because python3-tables isn't installed there, so the PyTables tests are skipped. Fix: Debian fixed this in 2.3.3+dfsg-4 (Debian #1145502) with debian/patches/hdf5_v2_compat.patch. Based on upstream pandas commit 389051d0, it marks the test as xfail(strict=False). We cherry-pick only that patch. Tested on stonking amd64 with python3-tables 3.11.1-3 and libhdf5-320 2.2.0+repack-5: the build succeeds and all test subsets pass. The io subset now reports 0 failed, 156 xfailed. ** Affects: pandas (Ubuntu) Importance: Undecided Status: New ** Tags: ftbfs update-excuse ** Tags added: ftbfs update-excuse -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/2169951 Title: pandas 2.3.3+dfsg-3ubuntu2 FTBFS: test_no_track_times fails with HDF5 2.x To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/pandas/+bug/2169951/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
