Hi, On 21/02/13 19:49, Jo Shields wrote: > Yeah, that looks pretty fucked. 20 failed JIT tests, including > pinvoke.exe (so calling into C libraries is broken) and jit-float.exe > (so floating point maths is broken). There's also a pretty low pass rate > on basic runtime tests. Compare to the 32-bit PPC tests: > > https://buildd.debian.org/status/fetch.php?pkg=mono&arch=powerpc&ver=2.10.8.1-7&stamp=1359380845 > > Overall results: tests: 16968, 100% pass, opt combinations: 21 > 399 test(s) passed. 0 test(s) did not pass. > > To the numbers from your build log: > > Overall results: tests: 16968, failed: 588, opt combinations: 21 (pass: > 96.53%) > 379 test(s) passed. 20 test(s) did not pass.
When it's important, consider making the build fail when tests fail. >> I remember that there was some mono process still running at 100% CPU >> _after_ build was finished that I needed to kill manually. But this was >> at a point where the build was already done and back to dpkg* already. >> Is this a known issue? > > It can happen on some arches where signal handling isn't working > properly Well, signal handling is working fine for thousands of other packages (and in other Linux projects on powerpcspe besides Debian), and manually killing (SIGTERM) worked fine, so I'm not convinced that the arch is the culprit. Roland -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
