Repository navigation
segfaults in test suite #3782
Description
I am getting segfaults from the test suite.
Program received signal SIGSEGV, Segmentation fault.
0x00007fffecadc34b in Triangulation::is_masked (this=this@entry=0x27e07e0, tri=tri@entry=1) at lib/matplotlib/tri/_tri.cpp:533
533 return !_mask.empty() && _mask(tri);
(gdb) bt
#0 0x00007fffecadc34b in Triangulation::is_masked (this=this@entry=0x27e07e0, tri=tri@entry=1) at lib/matplotlib/tri/_tri.cpp:533
#1 0x00007fffecae0c0c in Triangulation::calculate_neighbors (this=this@entry=0x27e07e0) at lib/matplotlib/tri/_tri.cpp:343
#2 0x00007fffecae1e46 in Triangulation::get_neighbors (this=0x27e07e0) at lib/matplotlib/tri/_tri.cpp:498
#3 0x00007fffecae7dbe in PyTriangulation_get_neighbors (self=<optimized out>, args=<optimized out>, kwds=<optimized out>)
at lib/matplotlib/tri/_tri_wrapper.cpp:142
#4 0x00007ffff7aefdb9 in call_function (oparg=<optimized out>, pp_stack=0x7ffffffc2bc8) at Python/ceval.c:4017
#5 PyEval_EvalFrameEx (f=<optimized out>, throwflag=<optimized out>) at Python/ceval.c:2679
[ -- snipped 1.6k (?!) calls in ceval.c -- ]
This is the backtrace from a 14.04 ubunutu box and I have seen a segfault on arch, but don't have a backtrace for it (yet).
Activity
tacaswell commentedon Nov 11, 2014
Travis does not reproduce this and a clean conda environment does not reproduce it. Closing as I assume that something went south with conda-related issues, sorry for the noise.
mdboom commentedon Nov 11, 2014
Could be the usual case of distutils not rebuilding enough... Wouldn't be surprised by this kind of thing at this point on master, of course.