Mailman 3 July 2003 - Cplusplus-sig

Problem with std::vector<boost::python::object>
by Pierre Barbier de Reuille July 30, 2003

July 30, 2003

I'd want to know why I can't compile this: void insert( vector<object>* o, object obj ) { insert_iterator<vector<object> > ii( *o, o->end() ); *ii = obj; } I use g++-3.2 and the output is: /usr/include/c++/3.2/bits/stl_vector.h:550: conversion from ` boost::python::api::object' to non-scalar type ` __gnu_cxx::__normal_iterator<boost::python::api::object*, std::vector<boost::python::api::object, std::allocator<boost::python::api::object> > >' requested Thansk, -- Pierre Barbier de Reuille INRA - UMR Cirad/Inra/Cnrs/Univ.MontpellierII AMAP Botanique et Bio-informatique de l'Architecture des Plantes TA40/PSII, Boulevard de la Lironde 34398 MONTPELLIER CEDEX 5, France tel : (33) 4 67 61 65 77 fax : (33) 4 67 61 56 68

2 2

Re: implititly_convertible<>() difficulty
by Jonathan Brandmeyer July 30, 2003

July 30, 2003

> To: c++-sig(a)python.org > From: David Abrahams <> > Date: Tue, 29 Jul 2003 15:44:12 -0400 > Subject: [C++-sig] Re: implititly_convertible<>() difficulty > Reply-To: c++-sig(a)python.org > > > I'm not sure what's going wrong here, but could you try this instead? > > #include <boost/python.hpp> > > class tuple_type > { > private: > double x; > double y; > double z; > public: > tuple_type( double a=0.0, double b=0.0, double c=0.0) > : x(a), y(b), z(c) {} > > tuple_type( boost::python::tuple sequence) > { > using boost::python::extract; > > int i = extract<int>( sequence.attr("__len__")); > switch (i) { > case 3: > z = extract<double>( sequence[2]); > // FALLTHROUGH > case 2: > y = extract<double>( sequence[1]); > x = extract<double>( sequence[0]); > break; > default: > break; > } > } > > tuple_type > operator+( const tuple_type& t) const > { return tuple_type( x+t.x, y+t.y, z+t.z); } > }; > > BOOST_PYTHON_MODULE( tuple_test) > { > using namespace boost::python; > > class_<tuple_type>( "tuple_type", init<double, double, double>()) > .def( self + self) > .def( self + other<tuple>()) > ; > } > > > -- > Dave Abrahams > Boost Consulting > www.boost-consulting.com > GCC rejected that with this error: tupletest.cpp: In function `void init_module_tuple_test()': tupletest.cpp:42: ambiguous overload for `boost::python::self_ns::self_t& + boost::python::other<boost::python::tuple>' operator c:/msys/local/include/boost/python/operators.hpp:184: candidates are: boost::python::detail::operator_<op_add, L, R> boost::python::self_ns::operator+(const L&, const R&) [with L = boost::python::self_ns::self_t, R = boost::python::other<boost::python::tuple>] c:/msys/local/include/boost/python/object_operators.hpp:60: boost::python::api::object boost::python::api::operator+(const L&, const R&) [with L = boost::python::self_ns::self_t, R = boost::python::other<boost::python::tuple>] Thanks, Jonathan Brandmeyer P.S. Please CC me in all replies on this thread.

2 1

Automatic downcasting (again)
by Stefano Delli Ponti July 30, 2003

July 30, 2003

I've read about a solution for automatic downcasting in threads from November to December 2002. Dave said he has commited some changes in the CVS. Are these changes going to be inserted in boost 1.30.1? Is it possible to know the affected files? TIA, Sted

2 3

implititly_convertible<>() difficulty
by Jonathan Brandmeyer July 29, 2003

July 29, 2003

I have a C++ type that I want to construct from a python tuple, and use python tuples in it's place. implicitly_convertible<Source, Target>() looked like the right thing, but it is not giving me the results that I had hoped. A complete example of what I am trying to do follows. This particular run was tested with GCC 3.2.2 (mingw) with Boost.Python 1.30.0 and Python 2.2.2 on WinXP. Thanks for you help, Jonathan Brandmeyer --------------------- tuple_test.cpp------------------------ #include <boost/python.hpp> class tuple_type { private: double x; double y; double z; public: tuple_type( double a=0.0, double b=0.0, double c=0.0) : x(a), y(b), z(c) {} tuple_type( boost::python::tuple sequence) { using boost::python::extract; int i = extract<int>( sequence.attr("__len__")); switch (i) { case 3: z = extract<double>( sequence[2]); // FALLTHROUGH case 2: y = extract<double>( sequence[1]); x = extract<double>( sequence[0]); break; default: break; } } tuple_type operator+( const tuple_type& t) const { return tuple_type( x+t.x, y+t.y, z+t.z); } }; BOOST_PYTHON_MODULE( tuple_test) { using namespace boost::python; class_<tuple_type>( "tuple_type", init<double, double, double>()) .def( self + self) ; implicitly_convertible<tuple, tuple_type>(); } ------------ and this is what I want to accomplish ---------------- Python 2.2.2 (#37, Oct 14 2002, 17:02:34) [MSC 32 bit (Intel)] on win32 Type "help", "copyright", "credits" or "license" for more information. >>> from tuple_test import tuple_type >>> x = tuple_type( 0.1, 3, 4) >>> y = (2, 3, 4) >>> z = tuple_type(y) Traceback (most recent call last): File "<stdin>", line 1, in ? TypeError: bad argument type for built-in operation >>> z = x+y Traceback (most recent call last): File "<stdin>", line 1, in ? TypeError: unsupported operand types for +: 'tuple_type' and 'tuple'

2 1

RE: [C++-sig] Re: Arbitrary argument lists, Python style?
by Conan Brink July 29, 2003

July 29, 2003

David Abrahams sez: > "Conan Brink" <conanb(a)STCG.net> writes: > > Forgive me if I'm just being dense [snip] > > > > class BackCaller: > > def SetCallback(self, action, *args): > > self.cb = (action, args) > > def CallItBack(self): > > if self.cb: > > (action, args) = self.cb > > return action(*args) > > > > The thing that's causing me to pull foot-long hairs from my head is that > > little *args > Does http://tinyurl.com/ibc6 help? It does somewhat, but not to the extent that I was hoping for. raw_function, of course, only wants an unbound function, not a member of a class. If I try to give it a class member, g++ 3.2 is more than pleased to give me the error messages you might expect about how one must use .* or ->* to call pointer-to-member function. Are there plans for a raw_method or something similar that I can use for wrapping a C++ class's member functions for this, or should I just go with the Python wrapper approach? Might be nice if the published version on www.boost.org included the documentation for raw_function.hpp (I get 404 errors for http://www.boost.org/libs/python/doc/v2/raw_function.html, which is linked from http://www.boost.org/libs/python/doc/v2/reference.html). -Conan This message contains information that may be confidential and privileged. Unless you are the addressee (or authorized to receive messages for the addressee), you may not use, copy, or disclose to anyone the message or any information contained in the message. If you have received the message in error, please advise the sender by reply e-mail and delete the message. Nothing in this message should be interpreted as a digital or electronic signature that can be used to authenticate a contract or any other legal document.

2 1

Virtual Functions with Default Implementations code sample
by Alex July 29, 2003

July 29, 2003

I can't build the "Virtual Functions with Default Implementations" code sample from the online tutorial : struct Base { virtual int f() { return 0; } }; struct BaseWrap : Base { BaseWrap(PyObject* self_) : self(self_) {} int f() { return call_method<int>(self, "f"); } int default_f() { return Base::f(); } // <<=== ***ADDED*** PyObject* self; }; ------- class_<Base, BaseWrap>("Base") .def("f", &Base::f, &BaseWrap::default_f) --> get the error "value_holder.hpp(131) : error C2661: 'BaseWrap::BaseWrap' : no overloaded function takes 2 parameters" ??? (btw I can build all the other code samples from the tutorial) ------- other newbie questions : Is there a place on the web where all the c++-sig@python mails are archived ? Does pyste can convert C++ function arguments to python keyword arguments ?

4 4

Arbitrary argument lists, Python style?
by Conan Brink July 28, 2003

July 28, 2003

Forgive me if I'm just being dense, but in hours of searching multiple archives, I have not been able to find a way to implement the following Python code in C++ via BPL (simplified somewhat for purposes of this discussion): class BackCaller: def SetCallback(self, action, *args): self.cb = (action, args) def CallItBack(self): if self.cb: (action, args) = self.cb return action(*args) The thing that's causing me to pull foot-long hairs from my head is that little *args Is such a thing possible, or do I need to do one of the following: 1) Tell my Python users that they have to explicitly wrap their args in a tuple? 2) Write a Python wrapper that does the translation back and forth between the variable arg list and a tuple that can be passed explicitly to and from my BPL-wrapped class? (and no, I'm not just doing this for fun - I've got some existing Python code that is being redone in C++ because it needs to live closer to the wire) -Conan This message contains information that may be confidential and privileged. Unless you are the addressee (or authorized to receive messages for the addressee), you may not use, copy, or disclose to anyone the message or any information contained in the message. If you have received the message in error, please advise the sender by reply e-mail and delete the message. Nothing in this message should be interpreted as a digital or electronic signature that can be used to authenticate a contract or any other legal document.

2 1

Segfault when calling python method from C++
by Pierre Barbier de Reuille July 28, 2003

July 28, 2003

I have an embedded python in a C++ program. I extend python using boost in order to interact with C++. Some of my methods makes use of C++ function-object. So I implemented this class to encapsulate Python methods : ***************************************************** struct PyCenterComp { PyCenterComp( boost::python::object o ); PyCenterComp( const PyCenterComp& copy ); vertex_type operator()( Delaunay2D::cell_descriptor cell ); boost::python::object method; }; PyCenterComp::PyCenterComp( object o ) : method( o ) { } PyCenterComp::PyCenterComp( const PyCenterComp& copy ) : method( copy.method ) { } vertex_type PyCenterComp::operator()( Delaunay2D::cell_descriptor cell ) { std::cerr << "Call method ... " << endl; boost::python::object o( method( cell_descriptor( cell ) ) ); std::cerr << "Called" << endl; extract<vertex_type> v( o ); std::cerr << "Converted" << endl; if( v.check() ) { return v(); } return vertex_type(); } ******************************************* The Delaunay2D::cell_descriptor being implemented as void*, it's encapsulated in a cell_descriptor structure which is exported in Python. In the python interpreter I created a function object with this code : ******************************************* class ProdMean: def __init__(self, walls): self.walls = walls def __call__(self,cell): poly = self.walls.getCell(cell) return self.mean(poly) def mean(self, poly): dx = 1 dy = 1 for i in poly: dx = dx * i.x() dy = dy * i.y() dx = dx ** (1.0/len(poly)) dy = dy ** (1.0/len(poly)) return vertex_type(dx,dy) ******************************************* Then I can use this class with : p = ProdMean(walls) l = [p(i) for i in walls] walls is a C++ object (a graph) and the getCell method returns a polygon (a vector of vertex). This little python code works well. But now, if I try to call p from C++ using the PyCenterComp object I get a segmentation fault when calling the method (I got this output: Call method ... zsh: segmentation fault ) If you want, I can give you the callstack when the crash occure, but it's quite huge (~50 elements ... ) so I'll do only if it's needed. Thanks, -- Pierre Barbier de Reuille INRA - UMR Cirad/Inra/Cnrs/Univ.MontpellierII AMAP Botanique et Bio-informatique de l'Architecture des Plantes TA40/PSII, Boulevard de la Lironde 34398 MONTPELLIER CEDEX 5, France tel : (33) 4 67 61 65 77 fax : (33) 4 67 61 56 68

2 11

Re: boost::python> exposing instances from c++
by David Abrahams July 28, 2003

July 28, 2003

The following message is a courtesy copy of an article that has been posted to comp.lang.python as well. Dear Mr. Handheld Vacuum, You will probably get better answers to your Boost.Python questions if you post them to the C++-sig. See http://www.boost.org/more/mailing_lists.htm#Cplussig. thedustbustr(a)aol.com (TheDustbustr) writes: > I'm writing a game in C++ which calls out to python for scripting. I'd like to > expose the instance of my Game class (a singleton) to python (so that the > instances of Clock, Camera, etc are available to the scripts). > > I ran into trouble trying to expose a specific instance (or even a function > returning the instance). Here is a simplification of my problem in code (c++): > > <code> > class A{ > public: > int i; > }; > class B{ > public: > A a; > }; > B b; > B getb() {return b;} //return the instance > > int main() { > getb().a.i = 42; > return 0;} > #include <boost/python/module.hpp> > #include <boost/python/class.hpp> > #include <boost/python/def.hpp> > using namespace boost::python; > BOOST_PYTHON_MODULE(hello) > { > class_<A>("A",init<>()) > .def("i", &A::i); > class_<B>("B",init<>()) > .def("a", &B::a); > def("getb", getb); > } > </code> > > Of course, I would be calling any python script using these classes from within > main(). > > This doesn't compile (using bjam with msvc). If I return by reference or > pointers in getb it gets ugly (like pages and pages of error messages). Am I > doing this the right way? How do I expose an instance to python? -- Dave Abrahams Boost Consulting www.boost-consulting.com

1 0

Qt class to embed python in a c++ application
by Pierre Barbier de Reuille July 28, 2003

July 28, 2003

Ok, here I send some files with classes to embed python. The use should be quite simple : you just have to instanciate the class PythonWnd and give it the module name and the module main function (which should be init#modulename). It should open another window in which you can enter python commands. It should even work without modules (but it's not tested). Thanks for your future comments :) -- Pierre Barbier de Reuille INRA - UMR Cirad/Inra/Cnrs/Univ.MontpellierII AMAP Botanique et Bio-informatique de l'Architecture des Plantes TA40/PSII, Boulevard de la Lironde 34398 MONTPELLIER CEDEX 5, France tel : (33) 4 67 61 65 77 fax : (33) 4 67 61 56 68

2 2