Mailman 3 March 2006 - Cplusplus-sig

C++ singleton
by Emmanuel Taurel Oct. 31, 2007

Oct. 31, 2007

Hello everybody, I am actually writing an interface to python for an existing C++ library using Boost. This library has a C++ singleton class. This class has only one constructor (with arguments) which is "protected" because it's a singleton. I am not able to build a Boost interface to this class because the compiler complains that the constructor I have given to Boost in the class_xxx constructor argument is not public (which is true) ! Therefore, my question is simple : How is it possible to interface such a class with only one constructor which is protected and a static method to create/access the object ? Thank's very much for your answer. Sincerely Emmanuel Taurel (etaurel(a)cells.es) PS : By the way, is there a searchable archive of this mailing list ?

4 3

optimizing away calls to the python runtime -- was [detecting if a function is written in python or C++]
by Mathieu Lacage May 15, 2006

May 15, 2006

hi, I apologize for this second email but the more I think about what I am trying to do, the less I see how to do it. So, here is a hopefully detailed explanation of what I am trying to achieve. Here are my c++ classes: class Event { virtual void notify (void) = 0; }; class Simulator { void schedule (Event *event); double now_s (void); // return current time. void run (void); }; sample c++ use code: class MyEvent : public Event { virtual void notify (void) { std::cout << "now=" Simulator::now_s () << std::endl; delete this; } }; Simulator->schedule (new MyEvent ()); Simulator->run (); // invokes MyEvent::notify sometime later. This sort of code requires a lot of "forwarding code". That is, you often have to write a bunch of Event-deriving classes whose sole purpose in life is to forward the call to their notify method to a call to another class' method with event-specific parameters. So, I have a small make_event c++ template which can be used to do this sort of thing: class MyClass { void random_stuff (double id) { std::cout << "id="<<id<<"now=" Simulator::now_s () << std::endl; } }; MyClass *obj = new MyClass (); Simulator->schedule (make_event (&MyClass::random_stuff, obj, 10.0)); Simulator->run (); Okay, so, to wrap this, I wrapped my Simulator class and my EventWrap class: struct EventWrap : yans::Event { EventWrap( PyObject* self_); virtual ~EventWrap(); virtual void notify (void); private: PyObject *m_self; }; EventWrap::EventWrap (PyObject* self_) : m_self(self_) { Py_INCREF(m_self); } EventWrap::~EventWrap() { Py_DECREF(m_self); } void EventWrap::notify (void) { call_method<void>(m_self, "notify"); delete this; } namespace boost { namespace python { template <> struct has_back_reference<EventWrap> : mpl::true_ {}; }} void export_event (void) { class_<EventWrap, std::auto_ptr<EventWrap>, boost::noncopyable> event ("Event"); event.def ("notify", pure_virtual(&Event::notify)); } void simu_insert_in_s (double delta, std::auto_ptr<EventWrap> ev) { Simulator::insert_in_s (delta, ev.get ()); ev.release (); } void export_simulator (void) { def ("run", Simulator::run); def ("insert_in_s", simu_insert_in_s); } which allows me to write the following python code: class __MyEvent__(Event): def set_callback (self, function, *args): self.__function_ = function; self.__args_ = args; def notify (self): self.__function_ (*self.__args_); def make_event(function, *args): ev = __MyEvent__ (); ev.set_callback (function, *args); return ev def my_nothing0 (): print "my model now=%f" % simulator.now_s (); simulator.insert_in_s (4.0, make_event (my_nothing1, 99)); simulator.run () okay, so all of this stuff works really nicely but there is a catch. I would like to be able to write code such as: simulator.insert_in_s (4.0, make_event (simulator.now_s)); simulator.run () and ensure that once simulator.run has reached the Simulator::run c++ method, no python code will be executed ever. Namely, I want call to make_event to generate a nice C++ Event class which will not call back into python's __MyEvent__.notify before calling simulator.now_s and Simulator::now_s. i.e., it should call back Simulator::now_s directly. The more I think about this, the less I see a way to do this with boost. Have I missed something ? Mathieu --

4 16

GCC version compatibility
by David Abrahams April 15, 2006

April 15, 2006

The following message is a courtesy copy of an article that has been posted to gmane.comp.python.devel as well. Recently people testing Boost.Python with GCC on Linux have reported that the extensions being tested have to be compiled with exactly the same version of GCC as the Python they're being loaded into, or they get mysterious crashes. That doesn't correspond to my past experience; it has always been true that, as long as the compiler used to build Python and the one used to build the extension have compatible 'C' ABIs, we've been okay. Yes, if you were going to pass types like FILE* across the Python/C API, then you additionally need to be sure that the two compilers are using the same 'C' library. That said, none of the Boost.Python tests do that. I'm wondering if there has been a well-known recent change either in Python or GCC that would account for these new reports. Any relevant information would be appreciated. -- Dave Abrahams Boost Consulting www.boost-consulting.com

8 36

RE: Wrapper for exception translation
by Gottfried.Ganssauge＠HAUFE.DE April 4, 2006

April 4, 2006

> -----Original Message----- > From: David Abrahams [mailto:dave@boost-consulting.com] > Sent: Monday, December 13, 2004 1:04 PM > To: c++-sig(a)python.org; Gottfried Ganßauge > Subject: Re: Wrapper for exception translation > > > Gottfried Ganßauge wrote: > > Hi Dave, > > > > I digged through the class_<> implementation and I think I > know what I > > must do to implement it. Like you suggested I would add another > > argument to the class_base constructor which is used as > another base > > class for the newly created class. > > Hi Gottfried, Hi Dave, > > I don't know if you're still listening, but today I offered to > implement this patch on the C++-sig (see my other post). > Are you still interested? I am still listening but was buried deeply with work so I even had to abandon the work already invested into this. Nonetheless: Yes - I'm still interested. Cheers, Gottfried P.S.: What second Post? I got 2 Copies of this one. > > > > -- > Dave Abrahams > Boost Consulting > http://www.boost-consulting.com >

5 9

How to deep-copy and delete a mixed python/C++ object from C++
by Achim H. April 3, 2006

April 3, 2006

Hi, I have a base class in C++ (MyClass), a wrapper class for boost::python (Wrapper) and a class derived from that in Python (MyPythonClass) that implements a pure virtual function of MyClass. So far so good. However, when I copy an object of Wrapper in C++, the new instance still points to the old python object and calls to the virtual function implemented in python will access the old data. I had hoped that the wrapper<MyClass> base class would copy the python object and everything would work. Furthermore, I want to be able to delete the object (in C++) through the pointer to its base class (it is stored in a std::vector<MyClass*>, together with MyClass objects completely implemented in C++). If I do now, the Python object stays around, still holds a pointer to a now dead object and at some point tries to kill it with not-so-nice consequences. How can I achieve both correct copying and deletion semantics ? I built a work-around: Wrapper is copied by a clone() function, that looks up the definition of the Python class (MyPythonClass) and, creates a new object from this class and then fills in the missing data (no data on the Python side, inheritance is only for overwriting virtual functions). However, then I get a lifetime problem with the object. Currently, I directly use Py_INCREF(copy) to keep it alive to the end of program runtime and then get an error message, beacuse the C++ object was already deleted when the std::vector<MyClass*> was cleaned up. TIA, Achim.

3 3

Problem with Templated class and Boost.python
by Arrakis Muadib March 31, 2006

March 31, 2006

Hi, I'm trying to create a binding for a small c++ class. I decided to created directly with Boost.Python and did'nt use Pyste cause it's not supported on my distro and I cannot make gccxml working. Now I'm confronted with a class having a template parameter. The signature is something like this: template<typename T> class A : virtual public B<T> { T methodX() {...} ... }; How can I map the template parameter to the boost.python class definition? Thank you Mathieu Leduc-Hamel

3 2

Re: [C++-sig] pyplusplus is alive :-)
by Niall Douglas March 31, 2006

March 31, 2006

On 29 Mar 2006 at 23:55, Roman Yakovenko wrote: > Problem number 1. pyplusplus generates invalid code mainly because of > unions that have unnamed struct inside. > > struct FXAPI Header > { > FXuint length; > char guid[16]; > union Flags > { > struct > { > FXuint hasItemIdList : 1; > FXuint pointsToFileOrDir : 1; > FXuint hasDescription : 1; > ... > }; > FXuint raw; > } flags; > } > > I am almost sure, that it can not be exported to Python. > First of all I am not sure how I should export unions? What do you > expect to get? The above is simply a technique for mapping bits to the same place as an int. Unnamed structs simply push their members into the enclosing namespace, therefore you'd access the above as header.flags.pointsToFileOrDir. Right now, pyplusplus is putting a $_x after flags because gccxml gives unnamed structs an internal name. To generate correct code, all pyplusplus needs to do is ignore any unnamed structs it sees. > According to this http://lists.boost.org/Archives/boost/2002/11/39157.php > it will not be supported. > > So I propose you to change union to struct, or to write some hand written code > that will export it. May be I could learn from it and will implement automatic > solution. But I don't promise. Thing is, we're not exporting a union. We're exporting the *members* of a union. Boost.Python does not need to know that the members refer to the same memory space - it just needs to know they are there like any other normal members of a struct. Worst case scenario - let pyplusplus only support accessing the first member of a union only. > Second and I think very important problem is unnamed structs within union. > In order to export bit fields pyplusplus creates wrapper that will > expose set\get > properties to the fields. In this case it do generates invalid code, because of > unnamed struct. > > What do you propose? pyplusplus should specifically filter out any unnamed structs it encounters and pretend they're not there. This will generate correct code. Cheers, Niall

2 3

library conflict
by Matthias G. Imhof March 30, 2006

March 30, 2006

I am using the C++API to embed python modules in a c++ code. I have a strange problem. The following module mkmodel works when called from interactive python, but fails when called from c++. from math import * import sys def replace(name, ext, num, args) : snum = str(num) ... narg = "EE"+str(n) arg = str(exp(args[j])) line = line.replace(narg, arg) narg = "TT"+str(n) arg = str(pow(10.0, args[j])) line = line.replace(narg, arg) ... def mkmodel(num, args) : replace("model", "rec", num, args) replace("model", "grd", num, args) ... return(0) def check() : print >>sys.stderr, "mkmodel check success" If either the math commands exp and pow, or these commands and the math import are removed, the c++ call works. I traced the problem with the debugger down to a problem with libmath.so. The c++ routine is listed below. The call to mkmodel.check() always succeeds independently of the math issue. But mkmodel.mkmodel(...) fails if mkmodel contains anything to do with libmath.so. (The PP commands are just simple wrappers combining a whole lot of PyXXX calls. They are taken from the Programming Python book. The situation remains the same when they are replaced with regular PyXXX routines.) const char cmd1[] = "import os,sys\nsys.path.append(os.getcwd())\n"; Py_Initialize(); PyRun_SimpleString(cmd1); if (PP_Run_Function("mkmodel", "check", "O", NULL, "()")) { printf("Can't call mkmodel check\n"); exit(0); } PyObject *tptr = PyTuple_New(*nd); for(int i=0; i<*nd; i++) { PyTuple_SetItem(tptr, i, Py_BuildValue("f", model[i])); } if (PP_Run_Function("mkmodel", "mkmodel", "O", NULL, "(iO)", modelcount, tptr)) { printf("Can't call mkmodel\n"); exit(0); } ps: I do link with -Xlinker --export-dynamic Matthias -- Matthias G.Imhof, Ph.D. Geosciences, Derring Hall 4044 (0420), Virginia Tech, Blacksburg, VA 24061 phone: (540) 231 6004, fax: (540) 231 3386, email: mgi(a)vt.edu, http://www.geol.vt.edu/profs/mgi -> Support Directive 10-289, it's not just another good idea ... <-

1 0

Wrappend objects identity problem
by Victor Nakoryakov March 30, 2006

March 30, 2006

Hi, I've got a problem with wrapping C++ classes using Boost.Python. From very beginning: I have some class UnsafePtr<T> that is holder for my classes. I have wrapped some class named CNavalWeaponSocket with construction: > python::class_< CNavalWeaponSocket, UnsafePtr<CNavalWeaponSocket>, boost::noncopyable >( "WeaponSocket", python::no_init ) > .def("getWeapon", &CNavalWeaponSocket::Weapon, python::return_value_policy<python::reference_existing_object>()); CNavalWeaponSocket::Weapon returns pointer (raw pointer, not UnsafePtr) to CWeaponObject (always the same), where CWeaponObject is class wrapped with: > python::class_< CWeaponObject, UnsafePtr<CWeaponObject>, boost::noncopyable >( "WeaponObject", python::no_init ); Consider following Python code: > # let 'c' is being instance of NavalWeaponSocket > print c.getWeapon() is c.getWeapon() > w = c.getWeapon() > w.focusWarder = c > print hasattr(c.getWeapon(), 'focusWarder') > print hasattr(w, 'focusWarder') Program output is: False False True But expected output for me is: True True True I.e. how much c.getWeapon() calls I make, so much copyes of that WeaponObject's __dict__, etc is made. The question is: how to acheive expected behaviour, and is it possible with BP at all? -- Victor (aka nail) Nakoryakov nail-mail<at>mail<dot>ru Krasnoznamensk, Moscow, Russia

5 15

Re: [C++-sig] Exporting an abstract class with virtual bool operator()(int)=0
by Achim H. March 29, 2006

March 29, 2006

Hi! IMHO, boost::python doesn't know that condition_Wrap is derived from condition, you'd have to mention that in the class_<>() statement. Achim. Pedro Crespo Valero wrote: > Hello everybody, > > I would like to export to python a function object that acts as a > predicate to specify "what > to filter" in my member function A::filter(pred). The point is that I > would like to define the predicates > in python so I decided to use an abstact class that I call "condition". > I include a little example below to clarify my problem. > Everything goes well and compiles but as you will see below I have > problems when I use it in python. > I would really appreciate if you could clarify me why is this happening > and some clue to solve it, > Thank you very much in advance for you help. > Pedro. > > > //////////////////////////////////////////// > > #include <boost/python.hpp> > #include <vector> > > // An abstract class > struct condition{ > virtual bool operator()(const int ii)=0; > virtual ~condition(){} > }; > > const int vals[]={1, 2, 3, 4, 5, 6, 7, 8, 9, 10}; > > class A{ > private: > std::vector<int> values_; > std::vector<int> fvalues_; > public: > A():values_(vals[0],vals[9]+1), fvalues_(0){} > > void filter(condition& pred) > { > fvalues_.resize(0); > for (std::size_t i=0; i<values_.size(); i++){ > if (pred(values_[i])) > fvalues_.push_back(values_[i]); > } > } > > }; > > //::::::::::::::::::::BOOST::PYTHON in action::::::::::::::::::::::::: > using namespace boost::python; > > struct condition_Wrap: condition, wrapper<condition>{ > bool operator()(const int ii){ return > call<bool>(this->get_override("operator()").ptr(),ii);} > }; > > BOOST_PYTHON_MODULE(test2py) > { > class_<condition_Wrap, boost::noncopyable>("condition") > .def("__call__", pure_virtual(&condition::operator())) > ; > > class_<A>("A") > .def("filter",&A::filter) > ; > } > > > In python:::::::::::::::::::::::::::::::::::::::::::::: > > from test2py import * > > # inherit condition and override __call__ > class less_than(condition): > def __init__(self,NoMx): > self.max=NoMx > > def __call__(self,n): > return (n<self.max) > > > x=A() > x.filter(less_than(5)) > > > And then BOOM!!!! the output is the following.... > > Traceback (most recent call last): > File "d:\My_Documents\pysource\test.py", line 19, in ? > x.filter(less_than(5)) > ArgumentError: Python argument types in > A.filter(A, less_than) > did not match C++ signature: > filter(class A {lvalue}, struct condition {lvalue}) > > > Thanks again!! > Long life to Boost.python !!! :-D >

3 4