from subprocess import call, PIPE, Popen import sys import re import pytest from numpy.testing import assert_ from numpy.compat import asbytes from scipy.linalg import _flapack as flapack # XXX: this is copied from numpy trunk. Can be removed when we will depend on # numpy 1.3 class FindDependenciesLdd: def __init__(self): self.cmd = ['ldd'] try: call(self.cmd, stdout=PIPE, stderr=PIPE) except OSError as e: raise RuntimeError("command %s cannot be run" % self.cmd) from e def get_dependencies(self, file): p = Popen(self.cmd + [file], stdout=PIPE, stderr=PIPE) stdout, stderr = p.communicate() if not (p.returncode == 0): raise RuntimeError("Failed to check dependencies for %s" % file) return stdout def grep_dependencies(self, file, deps): stdout = self.get_dependencies(file) rdeps = dict([(asbytes(dep), re.compile(asbytes(dep))) for dep in deps]) founds = [] for l in stdout.splitlines(): for k, v in rdeps.items(): if v.search(l): founds.append(k) return founds class TestF77Mismatch(object): @pytest.mark.skipif(not(sys.platform[:5] == 'linux'), reason="Skipping fortran compiler mismatch on non Linux platform") def test_lapack(self): f = FindDependenciesLdd() deps = f.grep_dependencies(flapack.__file__, ['libg2c', 'libgfortran']) assert_(not (len(deps) > 1), """Both g77 and gfortran runtimes linked in scipy.linalg.flapack ! This is likely to cause random crashes and wrong results. See numpy INSTALL.rst.txt for more information.""")