Mailman 3 January 2016 - Cplusplus-sig

Multiple registration safety and unit testing
by David Sankel April 21, 2016

April 21, 2016

Hello all, I'd like to allow a class to be registered more than once and wanted to get your opinion if this change would be a good idea. 'boost::python::class_' always adds a class to the current scope and registers it in the global registry. The global registry code (particularly the 'insert' function in 'src/converter/registry.cpp') asserts that the class hasn't been registered already. This, of course, prevents multiple registrations of the same type. There is a use case, though, where multiple registrations of the same type is both necessary and safe. Generally, a c++ component which has a 'class_' call may not know the name of the module it will eventually end up in. It would still be desirable to write a unit test for this component. If we write a unit test and put the 'class_' into some dummy module we run into a problem where other unit tests might call 'class_' in another scope. The safety of multiple 'class_' calls stems from always using the same conversion function. Semantically everything is a-okay. I'm proposing to change the precondition of the 'boost::python::converter::registry::insert' function from: The behavior is undefined unless the specified 'type_info' object has not already been registered. to: The behavior is undefined unless the specified 'type_info' object has not already been registered with a semantically different converter Any objections? -- David Sankel

2 3

passing NoneType as argument to constructor
by Liam Herron Jan. 29, 2016

Jan. 29, 2016

For the following code: ----------------------------------------------------------------------------------------------------------------------------------------------------------------------- #include <boost/python.hpp> using namespace boost::python; namespace // unnamed { class NullableDouble { public: NullableDouble() : isNull_(true) , value_(0.0) { } NullableDouble(double x) : isNull_(false) , value_(x) { } double value() const { return value_; // null check not relevant to this example } bool isNull() const { return isNull_; } // ... more functions but not needed for this example private: bool isNull_; double value_; }; } // end namespace unnamed void init_pnic() { class_<NullableDouble> ("NullableDouble", "NullableDouble", init<>()) .def(init<double>()) .def("value", &NullableDouble::value) .def("isNull", &NullableDouble::isNull) ; } BOOST_PYTHON_MODULE(_testPNIC) { init_pnic(); } ----------------------------------------------------------------------------------------------------------------------------------------------------------------------- I would like to have another constructor method in python that takes a 'None' and returns the same as the no arg constructor 'NullableDouble()' What is the best way to do this? Thanks, --Liam ============================================================================================= Email transmissions can not be guaranteed to be secure or error-free, as information could be intercepted, corrupted, lost, destroyed, arrive late or incomplete, or contain viruses. The sender therefore does not accept liability for any errors or omissions in the contents of this message which arise as a result of email transmission. In addition, the information contained in this email message is intended only for use of the individual or entity named above. If the reader of this message is not the intended recipient, or the employee or agent responsible to deliver it to the intended recipient, you are hereby notified that any dissemination, distribution,or copying of this communication, disclosure of the parties to it, or any action taken or omitted to be taken in reliance on it, is strictly prohibited, and may be unlawful. If you are not the intended recipient please delete this email message. ==============================================================================================

2 1

Re: [C++-sig] Passing memory allocated in C++ to Python
by Tony Cappellini Jan. 12, 2016

Jan. 12, 2016

Stefan, To: cplusplus-sig(a)python.org Subject: Re: [C++-sig] Passing memory allocated in C++ to Python Message-ID: <56945036.8080302(a)seefeld.name> Content-Type: text/plain; charset=windows-1252 > Essentially- Python needs access to the return value from a malloc() call. >>Really ? Why ? You say your C++ code needs to access the pointer(s). >>Python has no notion of "pointer", but it will happily pass around >>atever data you expose from C++. Sorry about that Stefan, my post should have been clearer. I know Python doesn't know about pointers. I should have said "Python needs access to the memory pointed to by the memory allocated with malloc()". >>f you expose the above class together with the constructor and thei >>'call_ioctl()' member function I think you have all you need. In your something class, the data type returned from allocate_memory() needs to be something that Python understands. Since that allocation function (member) will be allocating 100s of MB of memory, how will this memory map to a Python data type? Is a bytearray better to use than a list (as far as performance is concerned)? Is a list better to use than a string (as far as performance is concerned). Are there other data types that should be considered? The user will need to access that data as bytes, words, longs, and possibly sequences. I've tried using boost's extract, to convert the char * returned from malloc() into a Python string, and a Python list, but these result in compile errors.

2 4

Passing memory allocated in C++ to Python
by Tony Cappellini Jan. 12, 2016

Jan. 12, 2016

I've already got a project built for Linux, using Boost-python. My Python code can import the C++ extension and access the members that are exposed. My C++ code needs to call fill in C++ structures in order to call ioctl() functions, while passing the addresses of the structures to the ioctl functions. I also need to allocate large (10's of MB - 100s of MB) buffers which must be aligned to specific boundaries. A pointer to these buffers are part of the above-mentioned C++ structs. (I already have the C++ functions which handle the alignment and memory allocation) The problems I'm facing are: 1. How to initiate the memory allocation from Python, so that the user has access to the data after the ioctl() completes. This would eliminate the need from copying the data from C++ to Python, after each ioctl() call. 2. Using boost-python, which type of Python data structure is best suited to be shared across the C++/Python boundary? Keep in mind that the allocation needs to happen in C++ so that memory alignment can be maintained, but initiating the allocation/alignment will be done from Python. Essentially- Python needs access to the return value from a malloc() call. A link link to a working example of Python & C++ code would be helpful. . The boost wiki only has examples for exposing classes to Python. https://wiki.python.org/moin/boost.python/EmbeddingPython The examples I've found on StackOverflow aren't quite what I'm looking for. Thanks

2 1

Problem with shared_ptr converter in boost 1.60
by Carlos Roig Jan. 11, 2016

Jan. 11, 2016

Dear colleges, I work in a project which makes use of the Boost.python module. Las week we received feedback from some people that was having problems after updating boost from version 1.59 to 1.60. Specifically the problem appears while trying to use a shared_ptr class. Until now, we were registering our classes as follows: class_<Foo, Foo::Pointer, bases<>, boost::noncopyable >("FooClass"); being Foo::Pointer defined as boost::shared_ptr<Foo> Which automatically created the wrapper for python. Since version 1.60, the following error appears while trying to access that elements: >>> for Foo in Mesh.GetFooPointers() >>> print(Foo) TypeError: No to_python (by-value) converter found for C++ type: boost::shared_ptr<Foo> We have been "googling" for a while and we found a work-around which consist in adding: register_ptr_to_python<Foo::Pointer>(); This would be pretty annoying to do for all our code, as is quite big and was designed under the assumption of that boost::shared_ptr<> was automatically handled by boost.python and there was no need for calling "register_ptr_to_python", as specified here: https://wiki.python.org/moin/boost.python/PointersAndSmartPointers We haven't been able to identify any change that may be affecting this, so my question is: Is there any way in which we can restore the old behaviour? Does this change is inteended? If it helps, we have perfom tests using clang 3.7 and gcc 5.3, both with --std=c++11 enabled under linux64 (Kubuntu 15.10, Ubuntu 15.04 and 15.10, and Arch) using python 3.4.3 with identical results.. Best regards and thank you in advance, Charlie.

1 0

[Boost::Python] Registered class not recognized when returned by reference.
by Eugenio Bargiacchi Jan. 7, 2016

Jan. 7, 2016

Hello everyone, I have a member function with signature const Matrix2D & foo const (); Which returns an internal data structure. Matrix2D is using Matrix2D = Eigen::Matrix<double, Eigen::Dynamic, Eigen::Dynamic, Eigen::RowMajor | Eigen::AutoAlign>; I have explicitly set a to_python converter for Matrix2D, and I have exported the function with a call policy of return_internal_reference. However, when this function is called from Python I get TypeError: No Python class registered for C++ class Eigen::Matrix<double, -1, -1, 1, -1, -1> This happens also if the reference is not const. At the same time, if I modify the call policy to return_value_policy<copy_const_reference>() Python can decode the matrix just fine. The same holds true if I return the Matrix2D by value. However I really need to have a reference to the matrix in order store it into other classes. How can I make Python see the data structure through the reference? The converter in case it is needed: boost::python::list toPythonList(const double * data, size_t num) { boost::python::list list; for (size_t i = 0; i < num; ++i) list.append(data[i]); return list;} struct Matrix2DToPython{ static PyObject* convert(Matrix2D const & m) { boost::python::list list; for (int i = 0; i < m.rows(); ++i) { list.append(toPythonList(m.row(i).data(), m.cols())); } return boost::python::incref(list.ptr()); }}; Thanks, Eugenio

2 1

Beginner - PyObject_CallObject fails because imported module is not recognized
by BERGER Christian Jan. 6, 2016

Jan. 6, 2016

We have this simple (test) script that defines a function that we want to call via python's c api: import pprint import sys counter= 0 def onInitialize(ctxt): global counter counter+=1 pprint.pprint(sys.path) return counter While executing this works fine from the python command line, when using PyObject_CallObject to execute 'onInitialize', we're getting an error that pprint was undefined: "name 'pprint' is not defined". Moving the import statement into the function resolves the issue, but that seems wrong, and won't work for us (we want our customers to be able to provide their own implementation of onInitialize() and a bunch of other functions). I'm assuming I'm missing something very basic, but I haven't found it... The interesting thing is that the global variable counter is defined and increments as desired/expected when calling the function via PyObject_CallObject(.) repeatedly. The environment is Python 3.5.1 (64bit) Win7 The source code is below, if anybody wants the vs solution, that can be provided. Thanks for your help, Christian #include "stdafx.h" #include "Python.h" void hackIt() { const char* const pTestScripts = "import pprint\n" "import sys\n" "counter= 0\n" "def onInitialize(ctxt):\n" " global counter\n" " counter+=1\n" " pprint.pprint(sys.path)\n" " return counter\n"; Py_Initialize(); PyObject* pModule = PyImport_AddModule("__main__"); PyObject* pPyMainDict = PyModule_GetDict(pModule); _Py_static_string(PyId_string, "<string>"); PyObject* pFileName = _PyUnicode_FromId(&PyId_string); PyObject* pScriptObject = Py_CompileStringObject(pTestScripts, pFileName, Py_file_input, NULL, 0); PyObject* pLocalDictionary = PyDict_New(); if (pScriptObject) { PyObject* pResult = PyEval_EvalCode(pScriptObject, pPyMainDict, pLocalDictionary); if (NULL != pResult) { PyObject* pFxnObj = PyDict_GetItemString(pLocalDictionary, "onInitialize"); long retval = -1; PyObject* pTuple = PyTuple_New(1); PyObject* pParam = PyLong_FromLong(42); PyTuple_SetItem(pTuple, 0, pParam); //tuple takes ownership of pParam PyObject* pResultObj = PyObject_CallObject(pFxnObj, pTuple); Py_XDECREF(pTuple); if (pResultObj) { printf("got a %ld\n", PyLong_AsLong(pResultObj)); Py_DECREF(pResultObj); } else { PyObject* pPyErr = NULL; if (pPyErr = PyErr_Occurred()) { const char* pStr = NULL; PyObject* pTypeObj = NULL; PyObject* pValueObj = NULL; PyObject* pWhatObj = NULL; PyErr_Fetch(&pTypeObj, &pValueObj, &pWhatObj); if (pValueObj) { if (PyBytes_Check(pValueObj)) { printf("Error: %s\n", PyBytes_AsString(pValueObj)); } if (PyUnicode_Check(pValueObj)) { wchar_t buffer[1024]; PyUnicode_AsWideChar(pValueObj, buffer, 1024); printf("Error: %S\n", buffer); } } Py_XDECREF(pWhatObj); Py_XDECREF(pValueObj); Py_XDECREF(pTypeObj); } } } } Py_XDECREF(pLocalDictionary); Py_XDECREF(pScriptObject); Py_Finalize(); } int _tmain(int argc, _TCHAR* argv[]) { hackIt(); return 0; } This email and any attachments are intended solely for the use of the individual or entity to whom it is addressed and may be confidential and/or privileged. If you are not one of the named recipients or have received this email in error, (i) you should not read, disclose, or copy it, (ii) please notify sender of your receipt by reply email and delete this email and all attachments, (iii) Dassault Systemes does not accept or assume any liability or responsibility for any use of or reliance on this email. For other languages, go to http://www.3ds.com/terms/email-disclaimer

2 1