[Numpy-discussion] Numpy array flags - BUG?

Thomas Hrabe thrabe@gmail....
Thu Sep 11 16:04:57 CDT 2008


Hello everyone,

I must report odd behaviour of the numpy arrays regarding the flags set for
each array object in C++.
Please have a look at the following code:

static PyObject* test(PyObject* self,PyObject* args){

        int s[2];
        s[0] = 1;
        s[1] = 1;

        char* value = (char*)PyMem_Malloc(2*sizeof(int));

        PyObject* obj= PyArray_FromDimsAndData(1,s,NPY_INTLTR,(char*)value);
        PyArrayObject* array = (PyArrayObject*) obj;
        printf("%d\n",array->flags);
        array->flags=0;
        printf("%d\n",array->flags);
        return obj;
}


By creating my object with the code above, the flags of the object are not
set right.
The explicit setting of
array->flags=0;
would yield a print out of
0
for the next line, but when I look at the value from the interpreter, I
would get a
>>> a.flags.num
1
???
It seems to me the flags are overridden and that C alignment is a hard coded
default value.
I need to set the Flags to 1286, which means (Fortran alignment)

  C_CONTIGUOUS : False
  F_CONTIGUOUS : True
  OWNDATA : True
  WRITEABLE : True
  ALIGNED : True
  UPDATEIFCOPY : False

Thats exactly what I want to achieve in my real code(similar as above, but
more complex). But for that, however, I would get a
>>> a.flags.num
1285

  C_CONTIGUOUS : True
  F_CONTIGUOUS : False

  OWNDATA : True
  WRITEABLE : True
  ALIGNED : True
  UPDATEIFCOPY : False

even though I set the flags value in my C++ code to 1286 explicitly and a
printf directive prints 1286 .

Does anybody have a clue where the magic happens? Can I override the flags
settings of an array in python and set them to some value? I tried it with
the PyArray_UpdateFlags

function, but it did not work, same results.

Thank you in advance for your help,
Thomas 
-- 
View this message in context: http://www.nabble.com/Numpy-array-flags---BUG--tp19444726p19444726.html
Sent from the Numpy-discussion mailing list archive at Nabble.com.



More information about the Numpy-discussion mailing list