--pysparse --inline failures

[guyer]

There are at least six --inline failures (there appear to be more on OS X, plus it [segfaults before it finishes](issue #293))

==================================================================h4. 
FAIL: variable (examples.diffusion)
Doctest: examples.diffusion.variable
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/usr/lib/python2.7/doctest.py", line 2166, in runTest
raise self.failureException(self.format_failure(new.getvalue()))
AssertionError: Failed doctest test for examples.diffusion.variable
  File ".../examples/diffusion/variable.py", line 6, in variable

----------------------------------------------------------------------
File ".../examples/diffusion/variable.py", line 18, in examples.diffusion.variable
Failed example:
print var.allclose(values, atol = 1e-8, rtol = 1e-8)
Expected:
1
Got:
False


==============================================================h4. 
FAIL: howToWriteAScript (examples.levelSet.electroChem)
Doctest: examples.levelSet.electroChem.howToWriteAScript
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/usr/lib/python2.7/doctest.py", line 2166, in runTest
raise self.failureException(self.format_failure(new.getvalue()))
AssertionError: Failed doctest test for examples.levelSet.electroChem.howToWriteAScript
  File ".../examples/levelSet/electroChem/howToWriteAScript.py", line 6, in howToWriteAScript

----------------------------------------------------------------------
File ".../examples/levelSet/electroChem/howToWriteAScript.py", line 418, in examples.levelSet.electroChem.howToWriteAScript
Failed example:
print catalystVar.allclose(numerix.loadtxt(filepath), rtol=1e-4)
Expected:
1
Got:
False


==============================================================h4. 
FAIL: simpleTrenchSystem (examples.levelSet.electroChem)
Doctest: examples.levelSet.electroChem.simpleTrenchSystem
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/usr/lib/python2.7/doctest.py", line 2166, in runTest
raise self.failureException(self.format_failure(new.getvalue()))
AssertionError: Failed doctest test for examples.levelSet.electroChem.simpleTrenchSystem
  File ".../examples/levelSet/electroChem/simpleTrenchSystem.py", line 6, in simpleTrenchSystem

----------------------------------------------------------------------
File ".../examples/levelSet/electroChem/simpleTrenchSystem.py", line 25, in examples.levelSet.electroChem.simpleTrenchSystem
Failed example:
runSimpleTrenchSystem(numberOfSteps=2, displayViewers=False)
Exception raised:
Traceback (most recent call last):
File "/usr/lib/python2.7/doctest.py", line 1254, in __run
compileflags, 1) in test.globs
File "<doctest examples.levelSet.electroChem.simpleTrenchSystem[0]>", line 1, in <module>
runSimpleTrenchSystem(numberOfSteps=2, displayViewers=False)
File ".../examples/levelSet/electroChem/simpleTrenchSystem.py", line 314, in runSimpleTrenchSystem
bulkCatalystEquation.solve(bulkCatalystVar, dt = dt, solver=GeneralSolver(tolerance=1e-15, iterations=2000))
File ".../fipy/terms/term.py", line 215, in solve
solver._solve()
File ".../fipy/solvers/pysparse/pysparseSolver.py", line 108, in _solve
self._solve_(self.matrix, array, self.RHSvector)
File ".../fipy/solvers/pysparse/linearLUSolver.py", line 89, in _solve_
LU = superlu.factorize(L.matrix.to_csr())
RuntimeError: Factor is exactly singular
----------------------------------------------------------------------
File ".../examples/levelSet/electroChem/simpleTrenchSystem.py", line 37, in examples.levelSet.electroChem.simpleTrenchSystem
Failed example:
runSimpleTrenchSystem(numberOfSteps=2, catalystCoverage=0.1, displayViewers=False)
Exception raised:
Traceback (most recent call last):
File "/usr/lib/python2.7/doctest.py", line 1254, in __run
compileflags, 1) in test.globs
File "<doctest examples.levelSet.electroChem.simpleTrenchSystem[1]>", line 1, in <module>
runSimpleTrenchSystem(numberOfSteps=2, catalystCoverage=0.1, displayViewers=False)
File ".../examples/levelSet/electroChem/simpleTrenchSystem.py", line 314, in runSimpleTrenchSystem
bulkCatalystEquation.solve(bulkCatalystVar, dt = dt, solver=GeneralSolver(tolerance=1e-15, iterations=2000))
File ".../fipy/terms/term.py", line 215, in solve
solver._solve()
File ".../fipy/solvers/pysparse/pysparseSolver.py", line 108, in _solve
self._solve_(self.matrix, array, self.RHSvector)
File ".../fipy/solvers/pysparse/linearLUSolver.py", line 89, in _solve_
LU = superlu.factorize(L.matrix.to_csr())
RuntimeError: Factor is exactly singular


==============================================================h4. 
FAIL: gold (examples.levelSet.electroChem)
Doctest: examples.levelSet.electroChem.gold
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/usr/lib/python2.7/doctest.py", line 2166, in runTest
raise self.failureException(self.format_failure(new.getvalue()))
AssertionError: Failed doctest test for examples.levelSet.electroChem.gold
  File ".../examples/levelSet/electroChem/gold.py", line 6, in gold

----------------------------------------------------------------------
File ".../examples/levelSet/electroChem/gold.py", line 25, in examples.levelSet.electroChem.gold
Failed example:
runGold(numberOfSteps=10, displayViewers=False) # doctest: +GMSH
Exception raised:
Traceback (most recent call last):
File "/usr/lib/python2.7/doctest.py", line 1254, in __run
compileflags, 1) in test.globs
File "<doctest examples.levelSet.electroChem.gold[0]>", line 1, in <module>
runGold(numberOfSteps=10, displayViewers=False) # doctest: +GMSH
File ".../examples/levelSet/electroChem/gold.py", line 226, in runGold
advectionEquation.solve(distanceVar, dt = dt)
File ".../fipy/terms/term.py", line 215, in solve
solver._solve()
File ".../fipy/solvers/pysparse/pysparseSolver.py", line 108, in _solve
self._solve_(self.matrix, array, self.RHSvector)
File ".../fipy/solvers/pysparse/linearLUSolver.py", line 89, in _solve_
LU = superlu.factorize(L.matrix.to_csr())
RuntimeError: Factor is exactly singular


==============================================================h4. 
FAIL: leveler (examples.levelSet.electroChem)
Doctest: examples.levelSet.electroChem.leveler
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/usr/lib/python2.7/doctest.py", line 2166, in runTest
raise self.failureException(self.format_failure(new.getvalue()))
AssertionError: Failed doctest test for examples.levelSet.electroChem.leveler
  File ".../examples/levelSet/electroChem/leveler.py", line 6, in leveler

----------------------------------------------------------------------
File ".../examples/levelSet/electroChem/leveler.py", line 25, in examples.levelSet.electroChem.leveler
Failed example:
runLeveler(numberOfSteps=10, displayViewers=False, cellSize=0.25e-7) # doctest: +GMSH
Exception raised:
Traceback (most recent call last):
File "/usr/lib/python2.7/doctest.py", line 1254, in __run
compileflags, 1) in test.globs
File "<doctest examples.levelSet.electroChem.leveler[0]>", line 1, in <module>
runLeveler(numberOfSteps=10, displayViewers=False, cellSize=0.25e-7) # doctest: +GMSH
File ".../examples/levelSet/electroChem/leveler.py", line 419, in runLeveler
eqn.solve(var, boundaryConditions = BCs, dt = dt, solver=solver)
File ".../fipy/terms/term.py", line 215, in solve
solver._solve()
File ".../fipy/solvers/pysparse/pysparseSolver.py", line 108, in _solve
self._solve_(self.matrix, array, self.RHSvector)
File ".../fipy/solvers/pysparse/linearLUSolver.py", line 89, in _solve_
LU = superlu.factorize(L.matrix.to_csr())
RuntimeError: Factor is exactly singular


==================================================================
FAIL: stokesCavity (examples.flow)
Doctest: examples.flow.stokesCavity
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/usr/lib/python2.7/doctest.py", line 2166, in runTest
raise self.failureException(self.format_failure(new.getvalue()))
AssertionError: Failed doctest test for examples.flow.stokesCavity
  File ".../examples/flow/stokesCavity.py", line 6, in stokesCavity

----------------------------------------------------------------------
File ".../examples/flow/stokesCavity.py", line 294, in examples.flow.stokesCavity
Failed example:
print numerix.allclose(pressure.globalValue[...,-1], 162.790867927)
Expected:
1
Got:
False
----------------------------------------------------------------------
File ".../examples/flow/stokesCavity.py", line 296, in examples.flow.stokesCavity
Failed example:
print numerix.allclose(xVelocity.globalValue[...,-1], 0.265072740929)
Expected:
1
Got:
False
----------------------------------------------------------------------
File ".../examples/flow/stokesCavity.py", line 298, in examples.flow.stokesCavity
Failed example:
print numerix.allclose(yVelocity.globalValue[...,-1], -0.150290488304)
Expected:
1
Got:
False

Imported from trac ticket #395, created by guyer on 12-02-2011 at 18:47, last modified: 01-10-2012 at 13:07

[guyer]

Moved ~/.python27_compiled out of the way and ran PYTHONCOMPILED=MODULE python examples/diffusion/variable.py --inline obtaining

.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp: In function ‘PyObject* compiled_func(PyObject*, PyObject*)’:
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: error: ambiguous overload for ‘operator*’ in ‘py::object::operator[](int)(((j * ni) + i)) *#### pyobject::operator[](int)(((j * ni) + i))’
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: note: candidates are: operator*(int, int) <built-in>
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: note:                 operator*(int, float) <built-in>
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: note:                 operator*(int, double) <built-in>
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: note:                 operator*(float, int) <built-in>
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: note:                 operator*(float, float) <built-in>
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: note:                 operator*(float, double) <built-in>
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: note:                 operator*(double, int) <built-in>
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: note:                 operator*(double, float) <built-in>
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: note:                 operator*(double, double) <built-in>
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp: In function ‘PyObject* compiled_func(PyObject*, PyObject*)’:
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: error: ambiguous overload for ‘operator*’ in ‘py::object::operator[](int)(((j * ni) + i)) *#### pyobject::operator[](int)(((j * ni) + i))’
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: note: candidates are: operator*(int, int) <built-in>
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: note:                 operator*(int, float) <built-in>
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: note:                 operator*(int, double) <built-in>
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: note:                 operator*(float, int) <built-in>
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: note:                 operator*(float, float) <built-in>
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: note:                 operator*(float, double) <built-in>
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: note:                 operator*(double, int) <built-in>
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: note:                 operator*(double, float) <built-in>
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: note:                 operator*(double, double) <built-in>
Traceback (most recent call last):
  File "examples/diffusion/variable.py", line 65, in <module>
mesh = Tri2D(dx, dy, nx, ny)
  File ".../fipy/meshes/tri2D.py", line 100, in __init__
Mesh2D.__init__(self, vertices, faces, cells)
  File ".../fipy/meshes/mesh.py", line 84, in __init__
self._setGeometry(scaleLength = 1.)
  File ".../fipy/meshes/mesh.py", line 157, in _setGeometry
self._faceAreas = self._calcFaceAreas()
  File ".../fipy/meshes/mesh2D.py", line 79, in _calcFaceAreas
return numerix.sqrtDot(tangent, tangent)
  File ".../fipy/tools/numerix.py", line 421, in sqrtDot
""",result1=result1, a1=a1, a2=a2, ni=ni, NJ=NJ)
  File ".../fipy/tools/inline.py", line 111, in _runInline
extra_compile_args =['-O3'])
  File "/Users/guyer/.virtualenvs/virtualbrew/lib/python2.7/site-packages/scipy/weave/inline_tools.py", line 355, in inline
**kw)
  File "/Users/guyer/.virtualenvs/virtualbrew/lib/python2.7/site-packages/scipy/weave/inline_tools.py", line 482, in compile_function
verbose=verbose, **kw)
  File "/Users/guyer/.virtualenvs/virtualbrew/lib/python2.7/site-packages/scipy/weave/ext_tools.py", line 367, in compile
verbose = verbose, **kw)
  File "/Users/guyer/.virtualenvs/virtualbrew/lib/python2.7/site-packages/scipy/weave/build_tools.py", line 272, in build_extension
setup(name = module_name, ext_modules = [ext],verbose=verb)
  File "/Users/guyer/.virtualenvs/virtualbrew/lib/python2.7/site-packages/numpy/distutils/core.py", line 186, in setup
return old_setup(**new_attr)
  File "/Users/guyer/.homebrew/Cellar/python/2.7.2/lib/python2.7/distutils/core.py", line 169, in setup
raise SystemExit, "error: " + str(msg)
scipy.weave.build_tools.CompileError: error: Command "/usr/bin/g++-4.2 -fno-strict-aliasing -O3 -march=core2 -msse4.1 -w -pipe -DNDEBUG -g -fwrapv -O3 -Wall -I/Users/guyer/.virtualenvs/virtualbrew/lib/python2.7/site-packages/scipy/weave -I/Users/guyer/.virtualenvs/virtualbrew/lib/python2.7/site-packages/scipy/weave/scxx -I/Users/guyer/.virtualenvs/virtualbrew/lib/python2.7/site-packages/numpy/core/include -I/Users/guyer/.homebrew/Cellar/python/2.7.2/include/python2.7 -c .../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp -o .../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.o -O3" failed with exit status 1

I've run back to r3000 and still get the same error, so I think something has changed in SciPy (I'm running 0.9.0 now).

Trac comment by guyer on 12-03-2011 at 17:38

[guyer]

Replying to guyer:

Moved ~/.python27_compiled out of the way and ran PYTHONCOMPILED=MODULE python examples/diffusion/variable.py --inline obtaining

.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp: In function ‘PyObject* compiled_func(PyObject*, PyObject*)’:
.../fipy/tools/sc_8be0cdeed8d598d898ca8ab97055fc790.cpp:706: error: ambiguous overload for ‘operator*’ in ‘py::object::operator[](int)(((j * ni) + i)) *#### pyobject::operator[](int)(((j * ni) + i))’

Oops. That one I'd already figured out by patching sqrtDot:

#diff
--- fipy/tools/numerix.py   (revision 4996)
+++ fipy/tools/numerix.py   (working copy)
@@ -400,12 +400,22 @@
Usually used with v1==v2 to return magnitude of v1.
"""
unit1 = unit2 = 1
-        if _isPhysical(a1):
-            unit1 = a1.inBaseUnits().getUnit()
-            a1 = a1.numericValue
-        if _isPhysical(a2):
-            unit2 = a2.inBaseUnits().getUnit()
-            a2 = a2.numericValue
+        
+        def dimensionlessUnmasked(a):
+            unit = 1
+            mask = False
+            if _isPhysical(a):
+                unit = a.inBaseUnits().getUnit()
+                a = a.numericValue
+            if MA.isMaskedArray(a):
+                mask = a.mask
+                a = a.filled(fill_value=1)
+                
+            return (a, unit, mask)
+            
+        a1, unit1, mask1 = dimensionlessUnmasked(a1)
+        a2, unit2, mask2 = dimensionlessUnmasked(a2)
+        
NJ, ni = NUMERIX.shape(a1)
result1 = NUMERIX.zeros((ni,),'d')

@@ -420,6 +430,11 @@
result1[i] = sqrt(result1[i]);
""",result1=result1, a1=a1, a2=a2, ni=ni, NJ=NJ)

•    if NUMERIX.any(mask1) or NUMERIX.any(mask2):
  

•        result1 = MA.array(result1, mask=NUMERIX.logical_or(mask1, mask2))
  

•    print >>sys.stderr, "result1", type(result1)
  

  if unit1 != 1 or unit2 != 1:
  from fipy.tools.dimensions.physicalField import PhysicalField
  result1 = PhysicalField(value=result, unit=(unit1 * unit2)**0.5)

but then I get 
```python
.../fipy/tools/sc_32ebf90cbece946e767b6ffbe4bfa2100.cpp: In function ‘PyObject* compiled_func(PyObject*, PyObject*)’:
.../fipy/tools/sc_32ebf90cbece946e767b6ffbe4bfa2100.cpp:854: error: ‘normals_array’ was not declared in this scope
.../fipy/tools/sc_32ebf90cbece946e767b6ffbe4bfa2100.cpp: In function ‘PyObject* compiled_func(PyObject*, PyObject*)’:
.../fipy/tools/sc_32ebf90cbece946e767b6ffbe4bfa2100.cpp:854: error: ‘normals_array’ was not declared in this scope
Traceback (most recent call last):
  File "examples/diffusion/variable.py", line 86, in <module>
exec(fipy.tests.doctestPlus._getScript())
  File "<string>", line 6, in <module>
  File ".../fipy/terms/term.py", line 213, in solve
solver = self._prepareLinearSystem(var, solver, boundaryConditions, dt)
  File ".../fipy/terms/term.py", line 175, in _prepareLinearSystem
buildExplicitIfOther=self._buildExplcitIfOther)
  File ".../fipy/terms/unaryTerm.py", line 98, in _buildAndAddMatrices
diffusionGeomCoeff=diffusionGeomCoeff)
  File ".../fipy/terms/baseDiffusionTerm.py", line 334, in _buildMatrix
nthCoeffFaceGrad = coeff[numerix.newaxis] * var.faceGrad[:,numerix.newaxis]
  File ".../fipy/variables/variable.py", line 1555, in __getitem__
unit=self.unit,
  File ".../fipy/variables/variable.py", line 257, in _getUnit
return self._extractUnit(self.value)
  File ".../fipy/variables/variable.py", line 521, in _getValue
value = self._calcValue()
  File ".../fipy/variables/faceGradVariable.py", line 80, in _calcValue
return self._calcValueInline()
  File ".../fipy/variables/faceGradVariable.py", line 129, in _calcValueInline
shape=numerix.array(numerix.shape(tangents1)))
  File ".../fipy/tools/inline.py", line 185, in _runIterateElementInline
""")
  File "/Users/guyer/.virtualenvs/virtualbrew/lib/python2.7/site-packages/scipy/weave/inline_tools.py", line 355, in inline
**kw)
  File "/Users/guyer/.virtualenvs/virtualbrew/lib/python2.7/site-packages/scipy/weave/inline_tools.py", line 482, in compile_function
verbose=verbose, **kw)
  File "/Users/guyer/.virtualenvs/virtualbrew/lib/python2.7/site-packages/scipy/weave/ext_tools.py", line 367, in compile
verbose = verbose, **kw)
  File "/Users/guyer/.virtualenvs/virtualbrew/lib/python2.7/site-packages/scipy/weave/build_tools.py", line 272, in build_extension
setup(name = module_name, ext_modules = [ext],verbose=verb)
  File "/Users/guyer/.virtualenvs/virtualbrew/lib/python2.7/site-packages/numpy/distutils/core.py", line 186, in setup
return old_setup(**new_attr)
  File "/Users/guyer/.homebrew/Cellar/python/2.7.2/lib/python2.7/distutils/core.py", line 169, in setup
raise SystemExit, "error: " + str(msg)
scipy.weave.build_tools.CompileError: error: Command "/usr/bin/g++-4.2 -fno-strict-aliasing -O3 -march=core2 -msse4.1 -w -pipe -DNDEBUG -g -fwrapv -O3 -Wall -I/Users/guyer/.virtualenvs/virtualbrew/lib/python2.7/site-packages/scipy/weave -I/Users/guyer/.virtualenvs/virtualbrew/lib/python2.7/site-packages/scipy/weave/scxx -I/Users/guyer/.virtualenvs/virtualbrew/lib/python2.7/site-packages/numpy/core/include -I/Users/guyer/.homebrew/Cellar/python/2.7.2/include/python2.7 -c .../fipy/tools/sc_32ebf90cbece946e767b6ffbe4bfa2100.cpp -o .../fipy/tools/sc_32ebf90cbece946e767b6ffbe4bfa2100.o -O0" failed with exit status 1

and, indeed, normals is defined by

#cpp
py::object normals = convert_to_catchall(py_normals,"normals");

which seems to happen with MaskedArrays. Did weave stop handling those?

Trac comment by guyer on 12-03-2011 at 17:51

[wd15]

Inline issues have been fixed with changset:5095 and changset:5103.

Trac comment by wd15 on 01-10-2012 at 13:07