Tests fail on debian-mips

[olebole]

When compiling on debian-mips, python cpl fails with

make[3]: Leaving directory `/build/buildd-python-cpl_0.3.5.1-1-mips-oT7vkA/python-cpl-0.3.5.1/test/iiinstrumentp'
make[2]: Leaving directory `/build/buildd-python-cpl_0.3.5.1-1-mips-oT7vkA/python-cpl-0.3.5.1/test/iiinstrumentp'
for PYTHON in python2.7 python2.6; do \
        $PYTHON setup.py build_ext -i && \
        MALLOC_CHECK_=0 PYTHONPATH=$(pwd) $PYTHON test/TestRecipe.py \
            || exit $? ; \
    done
running build_ext
building 'cpl.CPL_recipe' extension
gcc -pthread -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -fPIC -I. -I/usr/include/python2.7 -c cpl/CPL_recipe.c -o build/temp.linux-mips-2.7/cpl/CPL_recipe.o
gcc -pthread -shared -Wl,-O1 -Wl,-Bsymbolic-functions build/temp.linux-mips-2.7/cpl/CPL_recipe.o -L. -lcplcore -lcpldfs -lcplui -lcpldrs -lgomp -lmcheck -o /build/buildd-python-cpl_0.3.5.1-1-mips-oT7vkA/python-cpl-0.3.5.1/cpl/CPL_recipe.so
....................*** glibc detected *** python2.7: double free or corruption (!prev): 0x00cfa0e0 ***
E: Caught signal 'Terminated': terminating immediately
make: *** [build] Terminated
make[1]: *** [override_dh_auto_test] Terminated

https://buildd.debian.org/status/logs.php?pkg=python-cpl&ver=0.3.5.1-1&arch=mips

SHA 2c3320394acb054e74f66f75daeb3cda3e030375

[olebole]

From the logfile, one sees that SIGTERM was emitted by the build script:

E: Caught signal 'Terminated': terminating immediately
make: *** [build] Terminated
make[1]: *** [override_dh_auto_test] Terminated
Build killed with signal TERM after 300 minutes of inactivity

This happened in test_crash:

self.recipe.param.crashing = 'free'
self.assertRaises(cpl.RecipeCrash, self.recipe, self.raw_frame)
self.recipe.param.crashing = 'segfault'
self.assertRaises(cpl.RecipeCrash, self.recipe, self.raw_frame)

The first test seemed to run until the recipe crashed (producing the * glibc * output), and either this did not return properly, or the second crash (segfault). So, notification of the parent process may not work properly here?

[olebole]

That seems no longer be the case, so closing this old bug.