On Tue, Feb 1, 2011 at 9:07 PM, Johan Hake <johan.h...@gmail.com> wrote:
> It looks like my "fix" did not help.
It did indeed fix the problem with "print v[int64(0)]". The correct
value (2.0) is printed out. The la unit tests reports a different
error (I don't know if it's related):
Testing basic PyDOLFIN linear algebra operations
------------------------------------------------
Running: MTL4Tester
Calling FFC just-in-time (JIT) compiler, this may take some time.
Calling FFC just-in-time (JIT) compiler, this may take some time.
EE..
Running: PETScTester
.Assertion failed: (dim <= 1), function size, file
/Users/fenicsslave/jhbuildbot/fenics-src/dolfin/dolfin/la/PETScBaseMatrix.h,
line 53.
[buildbotmac:87844] *** Process received signal ***
[buildbotmac:87844] Signal: Abort trap (6)
[buildbotmac:87844] Signal code: (0)
[buildbotmac:87844] [ 0] 2 libSystem.B.dylib
0x00007fff80f1235a _sigtramp + 26
[buildbotmac:87844] [ 1] 3 Python
0x0000000100088650 PyUnicodeUCS2_DecodeUnicodeEscape + 2608
[buildbotmac:87844] [ 2] 4 libSystem.B.dylib
0x00007fff80f8d9b4 __pthread_markcancel + 0
[buildbotmac:87844] [ 3] 5 libdolfin.0.dylib
0x00000001019d5689 _ZNK6dolfin11PETScMatrix4sizeEj + 121
[buildbotmac:87844] [ 4] 6 _cpp.so
0x0000000101583e7a _wrap_GenericTensor_size + 394
[buildbotmac:87844] [ 5] 7 Python
0x00000001000b88e3 PyEval_EvalFrameEx + 27571
[buildbotmac:87844] [ 6] 8 Python
0x00000001000b9af5 PyEval_EvalCodeEx + 2197
[buildbotmac:87844] [ 7] 9 Python
0x000000010003b4f5 function_call + 165
[buildbotmac:87844] [ 8] 10 Python
0x000000010000b902 PyObject_Call + 98
[buildbotmac:87844] [ 9] 11 Python
0x000000010001de1d instancemethod_call + 365
[buildbotmac:87844] [10] 12 Python
0x000000010000b902 PyObject_Call + 98
[buildbotmac:87844] [11] 13 Python
0x000000010007137b call_maybe + 315
[buildbotmac:87844] [12] 14 Python
0x000000010007308c slot_nb_multiply + 316
[buildbotmac:87844] [13] 15 Python
0x000000010000bbec binary_op1 + 156
[buildbotmac:87844] [14] 16 Python
0x000000010000e1d8 PyNumber_Multiply + 40
[buildbotmac:87844] [15] 17 Python
0x00000001000b404f PyEval_EvalFrameEx + 8991
[buildbotmac:87844] [16] 18 Python
0x00000001000b7eea PyEval_EvalFrameEx + 25018
[buildbotmac:87844] [17] 19 Python
0x00000001000b9af5 PyEval_EvalCodeEx + 2197
[buildbotmac:87844] [18] 20 Python
0x000000010003b5fd function_call + 429
[buildbotmac:87844] [19] 21 Python
0x000000010000b902 PyObject_Call + 98
[buildbotmac:87844] [20] 22 Python
0x00000001000b3ec2 PyEval_EvalFrameEx + 8594
[buildbotmac:87844] [21] 23 Python
0x00000001000b9af5 PyEval_EvalCodeEx + 2197
[buildbotmac:87844] [22] 24 Python
0x000000010003b4f5 function_call + 165
[buildbotmac:87844] [23] 25 Python
0x000000010000b902 PyObject_Call + 98
[buildbotmac:87844] [24] 26 Python
0x000000010001de1d instancemethod_call + 365
[buildbotmac:87844] [25] 27 Python
0x000000010000b902 PyObject_Call + 98
[buildbotmac:87844] [26] 28 Python
0x000000010007523a slot_tp_call + 74
[buildbotmac:87844] [27] 29 Python
0x000000010000b902 PyObject_Call + 98
[buildbotmac:87844] [28] 30 Python
0x00000001000b4fee PyEval_EvalFrameEx + 12990
[buildbotmac:87844] [29] 31 Python
0x00000001000b9af5 PyEval_EvalCodeEx + 2197
[buildbotmac:87844] *** End of error message ***
Abort trap
Johannes
> What we do is pretty simple and probaby error prone with different bit
> architechtures.
>
> from dolfin/swig/typemaps.i:
>
> bool Py_uint_convert(PyObject* in, dolfin::uint& value)
> {
> if (!PyInteger_Check(in))
> return false;
> long tmp = static_cast<long>(PyInt_AsLong(in));
> if (tmp<=0)
> return false;
> value = static_cast<dolfin::uint>(tmp);
> return true;
> }
>
> Here 'in' is the Python object that will be turned into an uint: value.
>
> I am happy for all suggestions on making thie typemap more robust.
>
> (I am not sure it is the typemap that causes the havoc but it sure looks like
> it. For example: a 0 or 1 is turned into: 70378776)
>
> Johan
>
> On Tuesday February 1 2011 09:16:19 Johan Hake wrote:
>> On Tuesday February 1 2011 08:38:02 Johannes Ring wrote:
>> > On Tue, Feb 1, 2011 at 5:28 PM, Johan Hake <johan.h...@gmail.com> wrote:
>> > > On Tuesday February 1 2011 04:02:03 Johannes Ring wrote:
>> > >> The problem can be triggered by the following code:
>> > >>
>> > >>
>> > >>
>> > >> import dolfin
>> > >>
>> > >> from numpy import int64
>> > >>
>> > >>
>> > >>
>> > >> v = dolfin.Vector(16)
>> > >>
>> > >>
>> > >>
>> > >> v[int(0)] = 2.0
>> > >>
>> > >> print v[int(0)]
>> > >>
>> > >>
>> > >>
>> > >> v[int64(0)] = 2.0
>> > >>
>> > >> print v[int64(0)]
>> > >>
>> > >>
>> > >>
>> > >> The output is:
>> > >>
>> > >>
>> > >>
>> > >> 2.0
>> > >>
>> > >> 0.0
>> > >>
>> > >>
>> > >>
>> > >> Can someone with access to a Mac try and see if they experience the
>> > >>
>> > >> same problem?
>> > >>
>> > >>
>> > >>
>> > >> Actually, I tried it now on my laptop (Lucid, 32 bit) and I got this
>> > >>
>> > >> error:
>> > >> >>> print v[int64(0)]
>> > >>
>> > >> Traceback (most recent call last):
>> > >>
>> > >> File "<stdin>", line 1, in <module>
>> > >>
>> > >> File "/tmp/local/lib/python2.6/site-packages/dolfin/cpp.py", line
>> > >>
>> > >> 3648, in __getitem__
>> > >>
>> > >> return _get_vector_single_item(self, indices)
>> > >>
>> > >> RuntimeError: index out of range
>> > >
>> > > It works on my 64 bit Maverick. I will see if I can find a sollution
>> > > for 32 bits. Any suggestions on the fly?
>> >
>> > No, I don't know.
>> >
>> > >> It works fine when I revert back to revision 5604 (Johan's original
>> > >>
>> > >> 5618 was lost in a merge).
>> > >
>> > > Do you suggest that the error came in the original 5618 changeset?
>> >
>> > Yes, the Mac buildbot was green before your commit with this comment:
>> >
>> > "Added a typemap macro for std::vector (passed by value) = for any
>> > Python sequence of scalars. = -- Enabled for value_shape, values
>> > (Constant) and coloring_type"
>>
>> Interesting. I changed a int/float => int/uint/double conversion, from my
>> own to using SWIGs built in type conversions. I have now changed back the
>> typemaps for int and uint. Let see what Mr Macbot says!
>> Johan
>>
>> > Build 1135 is green and build 1136 is red:
>> > http://fenicsproject.org:8080/builders/dolfin-mac-osx/builds/1136
>> > http://fenicsproject.org:8080/builders/dolfin-mac-osx/builds/1135
>> >
>> > Johannes
>> >
>> > > Johan
>> > >
>> > >> Johannes
>> > >>
>> > >> On Mon, Jan 31, 2011 at 10:19 AM, Johannes Ring <joha...@simula.no>
>>
>> wrote:
>> > >> > The Python la unit test fails on the Mac buildbot with the following
>> > >> >
>> > >> > error after revision 5618. Any ideas?
>> > >> >
>> > >> >
>> > >> >
>> > >> > Johannes
>> > >> >
>> > >> >
>> > >> >
>> > >> >
>> > >> >
>> > >> > Testing basic PyDOLFIN linear algebra operations
>> > >> >
>> > >> > ------------------------------------------------
>> > >> >
>> > >> >
>> > >> >
>> > >> > Running: MTL4Tester
>> > >> >
>> > >> > Calling FFC just-in-time (JIT) compiler, this may take some time.
>> > >> >
>> > >> > Calling FFC just-in-time (JIT) compiler, this may take some time.
>> > >> >
>> > >> > ..F.
>> > >> >
>> > >> > Running: PETScTester
>> > >> >
>> > >> > ..F.
>> > >> >
>> > >> > Running: uBLASDenseTester
>> > >> >
>> > >> > ..F.
>> > >> >
>> > >> > Running: uBLASSparseTester
>> > >> >
>> > >> > ..F.
>> > >> >
>> > >> > ====================================================================
>> > >> > ==
>> > >> >
>> > >> > FAIL: test_vector (__main__.MTL4Tester)
>> > >> >
>> > >> > --------------------------------------------------------------------
>> > >> > --
>> > >> >
>> > >> > Traceback (most recent call last):
>> > >> > File "test.py", line 162, in test_vector
>> > >> >
>> > >> > if org.owns_index(t(0)): self.assertAlmostEqual(org[t(0)], 2.0)
>> > >> >
>> > >> > AssertionError: 0.033607290958676872 != 2.0 within 7 places
>> > >> >
>> > >> >
>> > >> >
>> > >> > ====================================================================
>> > >> > ==
>> > >> >
>> > >> > FAIL: test_vector (__main__.PETScTester)
>> > >> >
>> > >> > --------------------------------------------------------------------
>> > >> > --
>> > >> >
>> > >> > Traceback (most recent call last):
>> > >> > File "test.py", line 162, in test_vector
>> > >> >
>> > >> > if org.owns_index(t(0)): self.assertAlmostEqual(org[t(0)], 2.0)
>> > >> >
>> > >> > AssertionError: 0.2551242083622477 != 2.0 within 7 places
>> > >> >
>> > >> >
>> > >> >
>> > >> > ====================================================================
>> > >> > ==
>> > >> >
>> > >> > FAIL: test_vector (__main__.uBLASDenseTester)
>> > >> >
>> > >> > --------------------------------------------------------------------
>> > >> > --
>> > >> >
>> > >> > Traceback (most recent call last):
>> > >> > File "test.py", line 162, in test_vector
>> > >> >
>> > >> > if org.owns_index(t(0)): self.assertAlmostEqual(org[t(0)], 2.0)
>> > >> >
>> > >> > AssertionError: 0.96740583429440807 != 2.0 within 7 places
>> > >> >
>> > >> >
>> > >> >
>> > >> > ====================================================================
>> > >> > ==
>> > >> >
>> > >> > FAIL: test_vector (__main__.uBLASSparseTester)
>> > >> >
>> > >> > --------------------------------------------------------------------
>> > >> > --
>> > >> >
>> > >> > Traceback (most recent call last):
>> > >> > File "test.py", line 162, in test_vector
>> > >> >
>> > >> > if org.owns_index(t(0)): self.assertAlmostEqual(org[t(0)], 2.0)
>> > >> >
>> > >> > AssertionError: 0.23277614769655719 != 2.0 within 7 places
>> > >> >
>> > >> >
>> > >> >
>> > >> > --------------------------------------------------------------------
>> > >> > --
>> > >> >
>> > >> > Ran 16 tests in 3.726s
>> > >> >
>> > >> >
>> > >> >
>> > >> > FAILED (failures=4)
>> > >> >
>> > >> > On Sun, Jan 30, 2011 at 4:00 AM, <build...@fenics.org> wrote:
>> > >> >> The Buildbot has detected a new failure of dolfin-mac-osx on FEniCS
>> > >> >>
>> > >> >> Buildbot. Full details are available at:
>> > >> >> http://fenicsproject.org:8080/builders/dolfin-mac-osx/builds/1136
>> > >> >>
>> > >> >> Buildbot URL: http://fenicsproject.org:8080/
>> > >> >>
>> > >> >>
>> > >> >>
>> > >> >> Buildslave for this Build: mac-osx
>> > >> >>
>> > >> >>
>> > >> >>
>> > >> >> Build Reason:
>> > >> >>
>> > >> >> Build Source Stamp: HEAD
>> > >> >>
>> > >> >> Blamelist: Johan Hake
>> > >> >>
>> > >> >>
>> > >> >>
>> > >> >> BUILD FAILED: failed dolfin check
>> > >> >>
>> > >> >>
>> > >> >>
>> > >> >> sincerely,
>> > >> >>
>> > >> >> -The Buildbot
>> > >> >>
>> > >> >> _______________________________________________
>> > >> >>
>> > >> >> Mailing list: https://launchpad.net/~dolfin
>> > >> >>
>> > >> >> Post to : dolfin@lists.launchpad.net
>> > >> >>
>> > >> >> Unsubscribe : https://launchpad.net/~dolfin
>> > >> >>
>> > >> >> More help : https://help.launchpad.net/ListHelp
>> > >>
>> > >> _______________________________________________
>> > >>
>> > >> Mailing list: https://launchpad.net/~dolfin
>> > >>
>> > >> Post to : dolfin@lists.launchpad.net
>> > >>
>> > >> Unsubscribe : https://launchpad.net/~dolfin
>> > >>
>> > >> More help : https://help.launchpad.net/ListHelp
>
_______________________________________________
Mailing list: https://launchpad.net/~dolfin
Post to : dolfin@lists.launchpad.net
Unsubscribe : https://launchpad.net/~dolfin
More help : https://help.launchpad.net/ListHelp
