On 03/08/2013 07:49 AM, Александров Петр wrote:
I have the C++ method which resize boost::numeric::array:void solve(const TReal start_t, const TReal end_t, boost::python::numeric::array &result) { ... result.resize(make_tuple(...)); ... } and this method is called from Python: some_object.solve(0, 10, solution) where solution = numpy.ndarray([]) In C++ file (where Python module is defined) I use numeric::array::set_module_and_type("numpy", "ndarray"); The program exits with the following message: Traceback (most recent call last): File ... some_object.solve(0, 10, solution) ValueError: cannot resize an array references or is referenced by another array in this way. Use the resize function How can I resize the boost::numeric::array?
I'm afraid it's probably impossible in this situation. NumPy arrays can share memory to create views, and because resizing requires the array to be reallocated, you're not allowed to do that if any views already exist - and that's impossible to guarantee if you don't control where your array is coming from. In general, I would recommend against trying resize NumPy arrays in *any* context. It just doesn't fit well with how arrays work under the hood, and IMO there should never have been a resize method. Jim _______________________________________________ Cplusplus-sig mailing list Cplusplus-sig@python.org http://mail.python.org/mailman/listinfo/cplusplus-sig
