Bug#1092638: openvdb: huge build log on s390x due to test failure spam, package still builds successfully
Philipp Kern <[email protected]>
| Newsgroups | gmane.linux.debian.ports.s390 |
|---|---|
| Organization | The Debian Project (https://www.debian.org) |
| Message-ID | <7b22b4b7-7509-4004-b124-fd938e124d41__41745.1608854876$1736505213$gmane$org@debian.org> |
Source: openvdb Version: 10.0.1-2.2 X-Debbugs-Cc: [email protected] openvdb built successfully on s390x but caused such a large build log that it could not be sent to the collector (because the transport is mail and it exceeded the mail size limit even when the log was compressed). I put it onto [0]. As far as I checked it mainly consists out of a single test failing and spamming the log continuously. The test failure was subsequently ignored. The failures seem to be precision failures like these: > 5: ./nanovdb/nanovdb/unittest/TestNanoVDB.cc:3434: Failure > 5: The difference between nanoAcc.getValue(p) and sphere(p) is 0.077666789293289185, which exceeds tolerance, where > 5: nanoAcc.getValue(p) evaluates to 0.02000001072883606, > 5: sphere(p) evaluates to -0.057666778564453125, and > 5: tolerance evaluates to 0.05000000074505806. > 6: ./nanovdb/nanovdb/unittest/TestOpenVDB.cc:1034: Failure > 6: The difference between approx and exact is 2.2453280687332153, which exceeds tolerance, where > 6: approx evaluates to -1.5676242113113403, > 6: exact evaluates to 0.677703857421875, and > 6: tolerance evaluates to 0.05000000074505806. Kind regards Philipp Kern [0] https://people.debian.org/~pkern/openvdb_10.0.1-2.2_s390x-2025-01-09T03:19:51Z.gz