Date: Sat, 16 Sep 2017 09:39:10 +0000 From: bugzilla-noreply@freebsd.org To: freebsd-toolchain@FreeBSD.org Subject: [Bug 221880] lang/gcc5: py-numpy/python2.7 core dump on FreeBSD 11.1-RELEASE amd64 Message-ID: <bug-221880-29464-TDD9GfQcWm@https.bugs.freebsd.org/bugzilla/> In-Reply-To: <bug-221880-29464@https.bugs.freebsd.org/bugzilla/> References: <bug-221880-29464@https.bugs.freebsd.org/bugzilla/>
next in thread | previous in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D221880 Martin Birgmeier <la5lbtyi@aon.at> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |la5lbtyi@aon.at --- Comment #3 from Martin Birgmeier <la5lbtyi@aon.at> --- I have a hunch that building py-numpy with gcc produces bad code which lead= s to coredumps. For example, since several months I cannot build py-matplotlib consistently any more - on most machines it fails, on one it succeeds. And graphics/qgis always dies with a signal 11. See bug #221622. And yes, I have already upgraded to gcc6, and it is the same - py-matplotlib cannot be recompiled except on one machine, and graphics/qgis (freshly recompiled) dies with a signal 11 after a few seconds. It would be interesting to have a py-numpy port which does not pull in gcc6= but rather uses the system's clang (which should be capable enough on 11.1). As a first try, running python2.7 setup.py build -j 4 install --prefix /tmp/z seems to compile o.k. (but not install due to complaints about PYTHONPATH). So, could the maintained of py-numpy create a version of the port which does not use gcc? -- Martin --=20 You are receiving this mail because: You are the assignee for the bug.=
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-221880-29464-TDD9GfQcWm>