Mailman 3 March 2006 - NumPy-Discussion

distutils.misc_util Configuration error
by Benjamin Thyreau 29 Mar '06

29 Mar '06

Hi, when trying to create an distutils.misc_util.Configuration instance, since numpy 0.9.5, i got an Python 2.4.1 (#1, May 2 2005, 15:06:50) [GCC 3.3.3 20040412 (Red Hat Linux 3.3.3-7)] on linux2 Type "help", "copyright", "credits" or "license" for more information. >>> import numpy.distutils.misc_util >>> numpy.distutils.misc_util.Configuration() Traceback (most recent call last): File "<stdin>", line 1, in ? File "/somewhere/numpy-0.9.6/build/lib.linux-i686-2.4/numpy/distutils/misc_util.py", line 450, in __init__ f = get_frame(i) File "/somewhere/numpy-0.9.6/build/lib.linux-i686-2.4/numpy/distutils/misc_util.py", line 359, in get_frame return sys._getframe(level+1) ValueError: call stack is not deep enough Error. Strangely enough (or not), when running from ipython, it seems to work. On jan 31th 2006, pearu commited: for i in range(1,3): f = get_frame(i) try: caller_instance = eval('self',f.f_globals,f.f_locals) break except NameError: caller_instance = None if isinstance(caller_instance, self.__class__): if caller_instance.options['delegate_options_to_subpackages']: self.set_options(**caller_instance.options) on misc_util.py, at the end of the Configuration constructor. I don't exactly understand what that code do, but it prevent my setup.py to run since then. Thanks for your help -- Benjamin Thyreau decideur.info

2 1

verbosity control for warnings
by Benjamin Georgi 29 Mar '06

29 Mar '06

Hi, my application frequently requires to take logarithms of zero, i.e. >>> import numarray >>> numarray.log(0.0) Warning: Encountered divide by zero(s) in log -inf Now, while the result of -inf works just fine, any direct application of numarray.log produces tons of warnings. Right now I'm masking the input arrays to avoid the warnings, which is rather inefficient. So, the question is whether there is some kind of verbosity control I could use to deactivate these warnings. If there is no such functionality, I would like to suggest looking into the build-in Python logging package for that purpose. Best, Benjamin

2 1

getting data pointer to numpy object in Pyrex
by Brian Blais 29 Mar '06

29 Mar '06

Hello, I posted this on the Pyrex list, but I figured someone here may be able to answer it too. Is here an easy way, in Pyrex, to get the data pointer to a numpy object which is *not* passed as a parameter? For example, I know I can do: #--------------- cdef extern from "numpy/arrayobject.h": struct PyArray_Descr: int type_num, elsize char type ctypedef class numpy.ArrayType [object PyArrayObject]: cdef char *data cdef int nd cdef int *dimensions, *strides cdef object base cdef PyArray_Descr *descr cdef int flags def test1(ArrayType w): # pass a numpy object cdef double *w_p # get the pointers w_p=<double *>w.data #--------------- ...but I would like to do something like... #------------------------------- def test2(d): # pass a python dict w=d['w'] # get the numpy object cdef double *w_p # get the pointers w_p=<double *>w.data #------------------------------- Is there a way to do this? thanks, Brian Blais -- ----------------- bblais(a)bryant.edu http://web.bryant.edu/~bblais

3 3

stringscalar comparisons with a chararray
by Michael Sorich 29 Mar '06

29 Mar '06

As shown in the example below, chararray comparisons do not seem to work well when stringscalars are used. In the example below I am also curious as to whether the the length of c[0] should be 2 instead of 5. numpy.__version__ 0.9.6 >>> from numpy import * >>> c =char.array(['on','two','three']) >>> c chararray([on, two, three], dtype='|S5') >>> c=='on' array([True, False, False], dtype=bool) >>> c==c[0] array([False, False, False], dtype=bool) >>> c==str(c[0]) array([True, False, False], dtype=bool) >>> c[0] 'on' >>> len(c[0]) 5 >>> c.dtype dtype('|S5') >>> c==c[2] array([False, False, True], dtype=bool)

1 0

Re: [Numpy-discussion] 64bit check_manyways segfault.
by Arnd Baecker 29 Mar '06

29 Mar '06

On Tue, 28 Mar 2006, Travis Oliphant wrote: > Arnd Baecker wrote: > > Hi, > > > > with current svn I get a segfault for numpy.test(10, 10): > > > > check_manyways (numpy.lib.tests.test_arraysetops.test_aso)Segmentation > > fault > > > Thanks for the test. Nobody had caught that one yet. > > Should be fixed now in SVN. That was quick, indeed! Works fine now. So presently the only problems on 64 Bit are the ndimage ones for scipy - I still have found no time (and won't during the next two weeks at least) to look into that in a bit more detail - sorry... Best, Arnd

1 0

Re: [Numpy-discussion] reshape and ravel methods of arrays now return views or raise error
by Travis Oliphant 29 Mar '06

29 Mar '06

Zachary Pincus wrote: >>> Unfortunately, the ravel/reshape methods seem to not be quite worked >>> out properly yet. Basically (and as expected) the ravel method >>> doesn't work anymore on fortran-strided arrays. But since >>> fortran-strided arrays happen *all-the-time* (e.g. after a transpose >>> operation), things get confusing fast. >>> > > > One question: was the issue in my example code where fortran-strided > arrays couldn't be reshaped at all a feature or a bug in the reshape > method? It's now a feature. Reshape will only work if it can produce a view of the array. Actually, I can't claim that I've thoroughly analyzed every possibility to determine if a view is possible. I'm sure there are cases where a view is technically feasible. The simple rules are: 1) If you are only adding or deleting ones from the shape, then it always succeeds. 2) If you aren't actually changing the shape it always succeeds (a -1 present means you are changing the shape) Other than that it only succeeds if you are reshaping a C-contiguous array using C-ordering or you are reshaping a Fortran-contiguous array using Fortran-ordering. -Travis

1 0

64bit check_manyways segfault.
by Arnd Baecker 28 Mar '06

28 Mar '06

Hi, with current svn I get a segfault for numpy.test(10, 10): check_manyways (numpy.lib.tests.test_arraysetops.test_aso)Segmentation fault There are no "incompatible pointer type" during the build of numpy. The backtrace is below. I hope this has not been reported during the last two days (I hardly manage to keep up with the traffic during the week ;-). Best, Arnd Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 46912507335168 (LWP 12055)] 0x00002aaaabaa4513 in PyArray_NewFromDescr (subtype=0x2aaaabbef320, descr=0x2aaaabbeba80, nd=1, dims=0x0, strides=0x0, data=0x0, flags=0, obj=0x0) at arrayobject.c:3969 3969 if (dims[i] < 0) { (gdb) bt #0 0x00002aaaabaa4513 in PyArray_NewFromDescr (subtype=0x2aaaabbef320, descr=0x2aaaabbeba80, nd=1, dims=0x0, strides=0x0, data=0x0, flags=0, obj=0x0) at arrayobject.c:3969 #1 0x00002aaaabaa6b2e in PyArray_Zeros (nd=1, dims=0x0, type=0x2aaaabad4c40, fortran_=0) at multiarraymodule.c:4679 #2 0x00002aaaabacebd5 in array_zeros (ignored=0x2aaaabbef320, args=0x2aaaabbef320, kwds=0x2aaaabad4c40) at multiarraymodule.c:4723 #3 0x0000000000478cfb in PyEval_EvalFrame (f=0x95dd30) at ceval.c:3558 #4 0x0000000000479fb1 in PyEval_EvalFrame (f=0x701c80) at ceval.c:3640 #5 0x0000000000479fb1 in PyEval_EvalFrame (f=0x716970) at ceval.c:3640 #6 0x0000000000479fb1 in PyEval_EvalFrame (f=0x7178c0) at ceval.c:3640 #7 0x0000000000479fb1 in PyEval_EvalFrame (f=0x6e4740) at ceval.c:3640 #8 0x000000000047ad2f in PyEval_EvalCodeEx (co=0x2aaaab948d50, globals=0x2aaaabad4c40, locals=0x0, args=0x6e4740, argcount=2, kws=0x91bad0, kwcount=0, defs=0x2aaaab950ce8, defcount=1, closure=0x0) at ceval.c:2736 #9 0x00000000004c6099 in function_call (func=0x2aaaab95e758, arg=0x2aaaae026758, kw=0x91b1e0) at funcobject.c:548 #10 0x0000000000417700 in PyObject_Call (func=0x2aaaabbef320, arg=0x2aaaabad4c40, kw=0x0) at abstract.c:1756 #11 0x00000000004772ea in PyEval_EvalFrame (f=0x6e7bb0) at ceval.c:3835 #12 0x000000000047ad2f in PyEval_EvalCodeEx (co=0x2aaaab948dc0, globals=0x2aaaabad4c40, locals=0x0, args=0x6e7bb0, argcount=2, kws=0x0, kwcount=0, defs=0x0, defcount=0, closure=0x0) at ceval.c:2736 #13 0x00000000004c6099 in function_call (func=0x2aaaab95e7d0, arg=0x2aaaae0265f0, kw=0x0) at funcobject.c:548 #14 0x0000000000417700 in PyObject_Call (func=0x2aaaabbef320, arg=0x2aaaabad4c40, kw=0x0) at abstract.c:1756 #15 0x0000000000420ee0 in instancemethod_call (func=0x2aaaabbef320, arg=0x2aaaae0265f0, kw=0x0) at classobject.c:2447 #16 0x0000000000417700 in PyObject_Call (func=0x2aaaabbef320, arg=0x2aaaabad4c40, kw=0x0) at abstract.c:1756 #17 0x00000000004777d9 in PyEval_EvalFrame (f=0x80c070) at ceval.c:3766 #18 0x000000000047ad2f in PyEval_EvalCodeEx (co=0x2aaaab93a340, globals=0x2aaaabad4c40, locals=0x0, args=0x80c070, argcount=2, kws=0x0, kwcount=0, defs=0x2aaaab9607a8, defcount=1, closure=0x0) at ceval.c:2736 #19 0x00000000004c6099 in function_call (func=0x2aaaab961cf8, arg=0x2aaaae026878, kw=0x0) at funcobject.c:548 #20 0x0000000000417700 in PyObject_Call (func=0x2aaaabbef320, arg=0x2aaaabad4c40, kw=0x0) at abstract.c:1756 #21 0x0000000000420ee0 in instancemethod_call (func=0x2aaaabbef320, arg=0x2aaaae026878, kw=0x0) at classobject.c:2447 #22 0x0000000000417700 in PyObject_Call (func=0x2aaaabbef320, arg=0x2aaaabad4c40, kw=0x0) at abstract.c:1756 #23 0x000000000044fd80 in slot_tp_call (self=0x2aaaad97d890, args=0x2aaaae04e450, kwds=0x0) at typeobject.c:4536 #24 0x0000000000417700 in PyObject_Call (func=0x2aaaabbef320, arg=0x2aaaabad4c40, kw=0x0) at abstract.c:1756 #25 0x00000000004777d9 in PyEval_EvalFrame (f=0x739bd0) at ceval.c:3766 #26 0x000000000047ad2f in PyEval_EvalCodeEx (co=0x2aaaab94c810, globals=0x2aaaabad4c40, locals=0x0, args=0x739bd0, argcount=2, kws=0x8af160, kwcount=0, defs=0x0, defcount=0, closure=0x0) at ceval.c:2736 #27 0x00000000004c6099 in function_call (func=0x2aaaab95f050, arg=0x2aaaae029248, kw=0x91afa0) at funcobject.c:548 #28 0x0000000000417700 in PyObject_Call (func=0x2aaaabbef320, arg=0x2aaaabad4c40, kw=0x0) at abstract.c:1756 #29 0x00000000004772ea in PyEval_EvalFrame (f=0x7301d0) at ceval.c:3835 #30 0x000000000047ad2f in PyEval_EvalCodeEx (co=0x2aaaab94c880, globals=0x2aaaabad4c40, locals=0x0, args=0x7301d0, argcount=2, kws=0x0, kwcount=0, defs=0x0, defcount=0, closure=0x0) at ceval.c:2736 #31 0x00000000004c6099 in function_call (func=0x2aaaab95f0c8, arg=0x2aaaae026d40, kw=0x0) at funcobject.c:548 #32 0x0000000000417700 in PyObject_Call (func=0x2aaaabbef320, arg=0x2aaaabad4c40, kw=0x0) at abstract.c:1756 #33 0x0000000000420ee0 in instancemethod_call (func=0x2aaaabbef320, arg=0x2aaaae026d40, kw=0x0) at classobject.c:2447 #34 0x0000000000417700 in PyObject_Call (func=0x2aaaabbef320, arg=0x2aaaabad4c40, kw=0x0) at abstract.c:1756 #35 0x000000000044fd80 in slot_tp_call (self=0x2aaaaca7acd0, args=0x2aaaae022c50, kwds=0x0) at typeobject.c:4536 #36 0x0000000000417700 in PyObject_Call (func=0x2aaaabbef320, arg=0x2aaaabad4c40, kw=0x0) at abstract.c:1756 #37 0x00000000004777d9 in PyEval_EvalFrame (f=0x7760a0) at ceval.c:3766 #38 0x0000000000479fb1 in PyEval_EvalFrame (f=0x6ec260) at ceval.c:3640 #39 0x000000000047ad2f in PyEval_EvalCodeEx (co=0x2aaaab93adc0, globals=0x2aaaabad4c40, locals=0x0, args=0x6ec260, argcount=3, kws=0x6d26d0, kwcount=0, defs=0x2aaaab963578, defcount=2, closure=0x0) at ceval.c:2736 #40 0x00000000004788f7 in PyEval_EvalFrame (f=0x6d2520) at ceval.c:3650 #41 0x000000000047ad2f in PyEval_EvalCodeEx (co=0x2aaaaab23dc0, globals=0x2aaaabad4c40, locals=0x0, args=0x6d2520, argcount=2, kws=0x66c900, kwcount=0, defs=0x2aaaaca84e30, defcount=2, closure=0x0) at ceval.c:2736 #42 0x00000000004788f7 in PyEval_EvalFrame (f=0x66c760) at ceval.c:3650 ---Type <return> to continue, or q <return> to quit--- #43 0x000000000047ad2f in PyEval_EvalCodeEx (co=0x2aaaab948030, globals=0x2aaaabad4c40, locals=0x0, args=0x66c760, argcount=0, kws=0x0, kwcount=0, defs=0x0, defcount=0, closure=0x0) at ceval.c:2736 #44 0x000000000047af72 in PyEval_EvalCode (co=0x2aaaabbef320, globals=0x2aaaabad4c40, locals=0x0) at ceval.c:484 #45 0x00000000004a1c72 in PyRun_InteractiveOneFlags (fp=0x2aaaaab132f0, filename=0x4cbf24 "<stdin>", flags=0x7fffff82fcfc) at pythonrun.c:1265 #46 0x00000000004a1e04 in PyRun_InteractiveLoopFlags (fp=0x2aaaab556b00, filename=0x4cbf24 "<stdin>", flags=0x7fffff82fcfc) at pythonrun.c:695 #47 0x00000000004a2350 in PyRun_AnyFileExFlags (fp=0x2aaaab556b00, filename=0x2aaaabad4c40 "number of dimensions must be >=0", closeit=0, flags=0x7fffff82fcfc) at pythonrun.c:658 #48 0x0000000000410788 in Py_Main (argc=0, argv=0x7fffff832721) at main.c:484 #49 0x00002aaaab34d5aa in __libc_start_main () from /lib64/tls/libc.so.6 #50 0x000000000040fdfa in _start () at start.S:113 #51 0x00007fffff82fdf8 in ?? () #52 0x00002aaaaabc19c0 in rtld_errno () from /lib64/ld-linux-x86-64.so.2 #53 0x0000000000000001 in ?? ()

1 0

Re: [Numpy-discussion] round_() problems
by Webb Sprague 28 Mar '06

28 Mar '06

> > > File "/var/www/localhost/htdocs/larry/LcUtil.py", line 120, in mat2table > > outStr += '' + str(N.round_(item,prec)) + '' > > > Have you tried to use around instead of round ? That way, I think you wouldn't > be using oldnumeric and stick to ndarrays... Yes. Same result.

1 0

NetCDF module that links against NumPy
by Travis Oliphant 28 Mar '06

28 Mar '06

I've adapted Konrad Hinsen's Scientific.IO.NetCDF module to work with NumPy. For now it's available at http://sourceforge.net/project/showfiles.php?group_id=1315&package_id=18550… I would like to put this module in SciPy but don't understand whether or not the License is compatible. It sounds GPL-ish. Would it be O.K. to place in the SciPy/sandbox? Regards, -Travis

2 1

segfault with min() method on fortran-strided arrays.
by Zachary Pincus 28 Mar '06

28 Mar '06

Hi, I found a segfault in numpy that seems to have to do with the new 'order' stuff. Steps to reproduce and gdb backtrace below. This is new since numpy 0.9.7.2262, which is what I updated from today. Zach In [1]: import numpy In [2]: a = numpy.array([[1,2],[3,4],[5,6]]) In [3]: b = numpy.transpose(a) In [4]: a.min() Out[4]: 1 In [5]: b.min() Program received signal EXC_BAD_ACCESS, Could not access memory. Reason: KERN_PROTECTION_FAILURE at address: 0x0000000c _check_axis (arr=0x76696577, axis=0xbfffeb0c, flags=0) at numpy/core/ src/arrayobject.c:3505 3505 *axis = PyArray_NDIM(temp)-1; (gdb) backtrace #0 _check_axis (arr=0x76696577, axis=0xbfffeb0c, flags=0) at numpy/ core/src/arrayobject.c:3505 #1 0x01170594 in array_min (self=0x6776b0, args=0x139eba0, kwds=0x0) at numpy/core/src/multiarraymodule.c:2909 #2 0x0028b984 in PyEval_EvalFrame (f=0x1868c10) at Python/ceval.c:3558 [...python stack frames snipped...] (gdb) info locals newdim = { ptr = 0xbfffeaa8, len = 1 } val = {-1} temp = (PyObject *) 0x6776b0 n = 2 (gdb) info args arr = (PyArrayObject *) 0x76696577 axis = (int *) 0xbfffeb0c flags = 0

1 1