Mailman 3 June 2008 - Cplusplus-sig

problem with boost::python::extract of c++ pointers from objects with multiple inheritance
by Carsten Harms June 25, 2008

June 25, 2008

Hi! I'm trying to extract a C++ pointer from a boost::python::object which inherited two base classes (both python extensions). I'm using boost 1.35 under MSVC 9. KeyListener exposing: > class_<KeyListenerWrap, boost::noncopyable>("KeyListener" ) > .def("keyPressed", > pure_virtual(&OIS::KeyListener::keyPressed), > return_value_policy<return_by_value>() ) > .def("keyReleased", > pure_virtual(&OIS::KeyListener::keyReleased), > return_value_policy<return_by_value>() ) > ; GameState exposing: > class_<GameStateBaseWrap, boost::noncopyable>("GameStateBase" ) > .def("enter", pure_virtual(&GameStateBase::enter)) > .def("exit", pure_virtual(&GameStateBase::exit)) > .def("pause", &GameStateBase::pause, &GameStateBaseWrap::pause) > .def("resume", &GameStateBase::resume, &GameStateBaseWrap::resume) > .def("frameStarted", > pure_virtual(&GameStateBaseWrap::frameStarted), > return_value_policy<return_by_value>() ) > .def("frameEnded", > pure_virtual(&GameStateBaseWrap::frameEnded), > return_value_policy<return_by_value>() ) > ; Please note that both c++ bases can be extracted as pointers when only _1_ is used as a base. for example: > class TestKeyListener(KeyListener): > def keyPressed(self, kevt): > print "key pressed:" > return True > > def keyReleased(self, kevt): > print "key released:" > return True > class TestState(GameStateBase): > def enter(self): > print "enter!" > def exit(self): > print "exit!" work as expected. I can extract the c++ pointer to their c++ base with > extract<OIS::KeyListener*> (_obj); or > extract<GameStateBase*> (_obj); But as soon as I inherit both like > class TestState2(GameStateBase, KeyListener): > def enter(self): > print "enter!" > def exit(self): > print "exit!" > > def keyPressed(self, kevt): > print "key pressed:" > return True > > def keyReleased(self, kevt): > print "key released:" > return True I can only extract the first base, GameStateBase in this example. If I exchange GameStateBase and KeyListener, I can only extract KeyListener. This is the error which I'm getting: > TypeError: No registered converter was able to extract a C++ pointer > to type class OIS::KeyListener from this Python object of type TestState2 Well, can someone tell me if I made a mistake, if this is a bug of boost::python, or if it just isn't implemented yet. Thanks in advance Carsten Harms

3 7

boost python pickle docs are empty on boost.org
by John Reid June 24, 2008

June 24, 2008

Does anyone else not get any content at http://www.boost.org/doc/libs/1_35_0/libs/python/doc/v2/pickle.html ? John.

3 3

Proposal to improve Python exception handling in Boost.Python
by Frank Benkstein June 24, 2008

June 24, 2008

Hi. Disclaimer: I'm quite new to Boost.Python, so I might have missed some things. If so, please tell me. Also I'm currently mostly concerned with embedding, not extending. When embedding Python with Boost.Python it is currently quite complicated to handle exceptions originating from Python code. If a Python error is detected by Boost.Python, error_already_set is thrown. Handling the original Python exception from C++ is cumbersome: <code> using namespace boost::python; try { object value = some_dict["key"]; ... } catch (error_already_set &) { if (PyErr_ExceptionMatches(PyExc_KeyError) { // Do actual exception handling ... } else { // We can't just rethrow the exception because that would terminate // the program. PyErr_Print(); ... } } </code> I would like to see at least the Python standard exceptions [1] to be translated to C++ exceptions by Boost.Python (possibly in the boost::python::exceptions namespace - mirroring the Python exceptions module) so the previous code could be written as follows: <code> using namespace boost::python; try { object value = some_dict["key"]; ... } catch (exceptions::key_error &e) { BOOST_ASSERT(e.args() == make_tuple("key")); // Do actual exception handling ... } catch (exceptions::base_exception &e) { std::cerr << e.traceback << std::flush; std::cerr << e << e.flush; } </code> Those exceptions could be automatically translated back to the original Python exceptions and the traceback restored (using PyErr_Restore). If you (the Boost.Python developers) agree to this proposal I would like to submit patches implementing the mentioned functionality through the following steps: 1. Convert all sites throwing error_already_set or calling throw_error_already_set to calls to handle_python_exception. handle_python_exception would just call throw_error_already_set itself. 2. Provide a registry for functions to be called when a specific Python exception is encountered. These functions get called with three boost::python::object arguments, the exception type, value and the traceback object when an exception is detected by handle_python_exception. As a fallback if no matching function is found error_already_set would still be thrown. 3. Provide a new namespace boost::python::exceptions with C++ exceptions that correspond to the Python standard exceptions[1] including inheritance. Each class should have a static boost::python::object member "mapped_type" wrapping the Python type object. And instance members "type", "value" and "traceback" so it can be set or restored as the Python error indicator using PyErr_Restore or PyErr_SetObject if the traceback is None. 4. Register all exceptions from 3. with the mechanism from 2. Remove the fallback throwing error_already_set from handle_python_exception because base_exception would match all unknown exceptions. Provide a runtime switch that would make handle_python_exception either use either the registry from 2. or just throw error_already_set. This is necessary to be compatible to code that is depending on error_already_set. I think it is safe to assume that this switch is guarded by the GIL and therefore thread safe. As you have probably noticed 3. and 4. will only work with Python >= 2.5. I don't think this is a problem, though, as there could be a simple compile switch around those features. Code that has to be compatible with older versions of python can still use the exception handling registry from 2. If need be 3. and 4. can also be adapted to work with older versions of Python but that is currently not important to me. Best regards, Frank Benkstein. [1] http://docs.python.org/api/standardExceptions.html -- Frank Benkstein Software Development Tel.: +49(0) 351 430093-27 Fax : +49(0) 351 430093-22 mailto:benkstein@langer-emv.de Langer EMV-Technik GmbH D-01728 Bannewitz Nöthnitzer Hang 31 Germany Registergericht: Amtsgericht Dresden, HRB 15402 Geschäftsführer: Gunter Langer www.langer-emv.de

2 5

Re: [C++-sig] FW: boost.python internal references on basic types
by Renaud Lepere June 18, 2008

June 18, 2008

> > class Foo > > { > > public: > > const int & get_x() { return m_x; } > > int m_x; > > }; > > BOOST_PYTHON_MODULE(hello) > > { > > class_<Foo>("Foo") > > .def("get_x", &Foo::get_x, > > return_internal_reference< >() ) > > ; > > } > > > > If I replace "int" with a proxy class containing the int > everything is > > ok ; but I would prefer not to change the c++ code. > > > > Can someone have an idea if what I try to do is possible ? > > Remove the "return_internal_reference". Trying to share int > references isn't something Python supports without adding > additional hand-crafted wrapper code. I replaced it with "return_value_policy<copy_const_reference>()" and it seems to be ok in this case. The other problem is for a non const function, that can be used to modify internals such as "int & x()", but i think i will not show it in python. Thanks!

1 0

FW: boost.python internal references on basic types
by English, Mark June 17, 2008

June 17, 2008

> -----Original Message----- > From: c++-sig-bounces(a)python.org > [mailto:c++-sig-bounces@python.org] On Behalf Of Renaud Lepere > Sent: 17 June 2008 13:52 > To: c++-sig(a)python.org > Subject: [C++-sig] boost.python internal references on basic types > > > Here is my simplified example > > class Foo > { > public: > void set_x(int & x) { m_x = x;} > const int & get_x() { return m_x; } > int m_x; > }; > BOOST_PYTHON_MODULE(hello) > { > class_<Foo>("Foo") > .def("set_x", &Foo::set_x) > .def("get_x", &Foo::get_x, > return_internal_reference< >() ) > ; > } > > If I replace "int" with a proxy class containing the int > everything is ok ; but I would prefer not to change the c++ code. > > Can someone have an idea if what I try to do is possible ? Remove the "return_internal_reference". Trying to share int references isn't something Python supports without adding additional hand-crafted wrapper code. _______________________________________________________________________ This email is intended only for the use of the individual(s) to whom it is addressed and may be privileged and confidential. Unauthorised use or disclosure is prohibited. If you receive this e-mail in error, please advise immediately and delete the original message without copying, using, or telling anyone about its contents. This message may have been altered without your or our knowledge and the sender does not accept any liability for any errors or omissions in the message. This message does not create or change any contract. Royal Bank of Canada and its subsidiaries accept no responsibility for damage caused by any viruses contained in this email or its attachments. Emails may be monitored. RBC Capital Markets is a business name used by branches and subsidiaries of Royal Bank of Canada, including Royal Bank of Canada, London branch and Royal Bank of Canada Europe Limited. In accordance with English law requirements, details regarding Royal Bank of Canada Europe Limited are set out below: ROYAL BANK OF CANADA EUROPE LIMITED Registered in England and Wales 995939 Registered Address: 71 Queen Victoria Street, London, EC4V 4DE. Authorised and regulated by the Financial Service Authority. Member of the London Stock Exchange

1 0

boost.python internal references on basic types
by Renaud Lepere June 17, 2008

June 17, 2008

I would like to wrap member functions returning internal references on int / double. I found an example similar to my problem but not exactly the same on http://www.boost.org/doc/libs/1_35_0/libs/python/doc/v2/return_internal_ reference.html Here is my simplified example class Foo { public: void set_x(int & x) { m_x = x;} const int & get_x() { return m_x; } int m_x; }; BOOST_PYTHON_MODULE(hello) { class_<Foo>("Foo") .def("set_x", &Foo::set_x) .def("get_x", &Foo::get_x, return_internal_reference< >() ) ; } this leads me to the given error ..\boost/python/object/make_instance.hpp(24) : error C2027: utilisation du type non defini 'boost::STATIC_ASSERTION_FAILURE<x>' this corresponds to BOOST_STATIC_ASSERT(is_class<T>::value); If i replace "int" with a proxy class containing the int evrything is ok ; but i would prefer not to change the c++ code. Can someone have an idea if what i try to do is possible ? Thanks Renaud

2 1

Re: [C++-sig] shared_ptr<> to Boost.Python wrapped object going out ofscope early?
by English, Mark June 16, 2008

June 16, 2008

> -----Original Message----- > From: c++-sig-bounces(a)python.org > [mailto:c++-sig-bounces@python.org] On Behalf Of Christopher Wicklein > Sent: 16 June 2008 16:43 > > Python code above. To fill things out a bit, here is the > Boost.Python statement used to wrap the Observer: > > class_<ObserverWrap, boost::noncopyable>("Observer") > .def("onModify", pure_virtual(&Observer::onModify)) > ; > > Here is the wrapper class: > > struct ObserverWrap : Observer, wrapper<Observer> { > onModify(Record record) { > this->get_override("onModify")(record); } } > What is the function signature of "onModify()" ? Should the function signature be based around shared_ptr (I've also included a void return type) ? E.g. struct ObserverWrap : Observer, wrapper<Observer> { void onModify(shared_ptr<Record> record) { this->get_override("onModify")(record); } } _______________________________________________________________________ This email is intended only for the use of the individual(s) to whom it is addressed and may be privileged and confidential. Unauthorised use or disclosure is prohibited. If you receive this e-mail in error, please advise immediately and delete the original message without copying, using, or telling anyone about its contents. This message may have been altered without your or our knowledge and the sender does not accept any liability for any errors or omissions in the message. This message does not create or change any contract. Royal Bank of Canada and its subsidiaries accept no responsibility for damage caused by any viruses contained in this email or its attachments. Emails may be monitored. RBC Capital Markets is a business name used by branches and subsidiaries of Royal Bank of Canada, including Royal Bank of Canada, London branch and Royal Bank of Canada Europe Limited. In accordance with English law requirements, details regarding Royal Bank of Canada Europe Limited are set out below: ROYAL BANK OF CANADA EUROPE LIMITED Registered in England and Wales 995939 Registered Address: 71 Queen Victoria Street, London, EC4V 4DE. Authorised and regulated by the Financial Service Authority. Member of the London Stock Exchange _______________________________________________ C++-sig mailing list C++-sig(a)python.org http://mail.python.org/mailman/listinfo/c++-sig

3 2

shared_ptr<> to Boost.Python wrapped object going out of scope early?
by Christopher Wicklein June 16, 2008

June 16, 2008

I've been having problems getting a C++ class wrapped with Boost.Python to be shared appropriately between Python and C++ library code. The Python program looks like this: class Watcher(Observer): def __init__(self): Observer.__init__(self) def onModify(self, record): print record database = Database() watcher = Watcher() database.watch(some_key, watcher) cl_service_all() The basic idea is that an instance of a wrapped database record observer is created, the observer is registered with the C++ object wrapped by database via a passed shared_ptr (the second argument of database.watch) and retained by the library as a weak_ptr, and the program drops into a C-based event loop which is integrated with the C+ + library providing the observer and database classes. The C++ library fails at runtime because the weak_ptr becomes invalid. It seems that the shared_ptr passed to database.watch isn't retained by the Python code above. To fill things out a bit, here is the Boost.Python statement used to wrap the Observer: class_<ObserverWrap, boost::noncopyable>("Observer") .def("onModify", pure_virtual(&Observer::onModify)) ; Here is the wrapper class: struct ObserverWrap : Observer, wrapper<Observer> { onModify(Record record) { this->get_override("onModify")(record); } } Having done a little diagnostic work by printf, I believe that an instance of Oberver is created, passes as a shared_ptr to the library, and accessible to the library prior to entering the event loop. When the weak pointer which was good before entering the event loop is locked from within the event loop, the weak pointer has become invalid. I presume that what I'm trying to do (hold a wrapped object by a shared_ptr on the Python side and allow a C++ library to hold a weak_ptr to that object) isn't too strange and that there should be some idiomatic way to make this work. Is there something in particular I need to do so that the Python side doesn't drop the shared pointer, or is it clear from the code above that I have some critical misunderstanding of how to use Boost.Python? ---- Christopher Wicklein <chrisw(a)wicklein.org>

1 0

Re: [C++-sig] Help with problem caused going from boost_1_33_1 to 1_35_0
by Ralf W. Grosse-Kunstleve June 13, 2008

June 13, 2008

I stared at your code for a few minutes but I don't have an answer for you. It looks like it should work. I'm not aware of changes in Boost.Python that could make your code fail. I'd probably try to reverse the import order: from spucb.iir_coeff import * from spucb.butterworth_iir import * although I don't expect this to make a difference. Ralf ----- Original Message ---- From: Tony Kirke <pyspuc(a)gmail.com> To: c++-sig(a)python.org Sent: Monday, June 9, 2008 8:25:20 AM Subject: [C++-sig] Help with problem caused going from boost_1_33_1 to 1_35_0 Hi, My c++ code that worked fine for Boost_1_33_1 (building with v1 system) is now giving the following error with 1_35_0: Boost.Python.ArgumentError: Python argument types in spucb.butterworth_iir.butterworth_iir(iir_coeff, float) did not match C++ signature: butterworth_iir(iir_coeff {lvalue}, float) I done a lot of searching to find out what changed in the newer Boost/boost build system and could not figure out what to do. Below is all of the code for a simple testcase. Since I have lots of code that is broken in this way, I'm wondering if there is an easy change that can be made to the boost wrapping code (or source itself) to make this functional. Is there a simple example of how this problem is solved. Was there a fundamental reason this is no longer supported? Thanks Tony ------------------------------------------ class iir_coeff { public: long order; public: iir_coeff(long ord=1); ~iir_coeff(); }; ------------------------------------------ #include "iir_coeff.h" iir_coeff::iir_coeff(long ord) { order = ord; } iir_coeff::~iir_coeff() {} ------------------------------------------ // Boost Includes ============================================================== #include <boost/python.hpp> #include <boost/cstdint.hpp> // Includes ==================================================================== #include "iir_coeff.h" // Using ======================================================================= using namespace boost::python; // Module ====================================================================== BOOST_PYTHON_MODULE(iir_coeff) { class_< iir_coeff >("iir_coeff", init< const iir_coeff& >()) .def(init< optional< long int > >()) .def_readwrite("order", &iir_coeff::order) ; } ------------------------------------------ #include "iir_coeff.h" void butterworth_iir(iir_coeff& filt, float fcd); ------------------------------------------ #include "butterworth_iir.h" void butterworth_iir(iir_coeff& filt, float fcd) { long order = filt.order; filt.order = order+1; } ------------------------------------------ // Boost Includes ============================================================== #include <boost/python.hpp> #include <boost/cstdint.hpp> // Includes ==================================================================== #include "butterworth_iir.h" // Using ======================================================================= using namespace boost::python; // Module ====================================================================== BOOST_PYTHON_MODULE(butterworth_iir) { def("butterworth_iir", &butterworth_iir); } ------------------------------------------ use-project boost : ../../boost_1_35_0 ; project : requirements <library>/boost/python//boost_python/<link>static : usage-requirements <include>../../boost_1_35_0 <include>. ; lib spuc : butterworth_iir.cpp iir_coeff.cpp : <include>. <include>../../boost_1_35_0 ; alias staticspuc : spuc/<link>static ; import python ; python-extension butterworth_iir : py_butterworth_iir.cpp : <include>. <include>../../boost_1_35_0 : <library>staticspuc ; python-extension iir_coeff : py_iir_coeff.cpp : <include>. <include>../../boost_1_35_0 : <library>staticspuc ; install dist : __init__.py butterworth_iir iir_coeff : <location>/usr/lib/python2.4/site-packages/spucb ; ------------------------------------------ boost-build ../../boost_1_35_0/tools/build/v2 ; ------------------------------------------ from spucb.butterworth_iir import * from spucb.iir_coeff import * x=iir_coeff(8) butterworth_iir(x,0.24) print x.order ------------------------------------------ Traceback (most recent call last): File "tes.py", line 4, in ? butterworth_iir(x,0.24) Boost.Python.ArgumentError: Python argument types in spucb.butterworth_iir.butterworth_iir(iir_coeff, float) did not match C++ signature: butterworth_iir(iir_coeff {lvalue}, float)

3 7

ED MATRIX - A magazine for Education & Development
by ed＠edmatrix.net June 13, 2008

June 13, 2008

<http://www.edmatrix.net> Subscription Details : Year Offer Rates Subscription Rates 1 year 720/- 900/- 2 years 1350/- 1800/- 3 years 1890/- 2700/- For more details Contact : 9790977297, 044 - 25569288 e-mail : ed(a)edmatrix.net <mailto:%20ed@edmatrix.net> , edmatrix(a)edmatrix.net website : www.edmatrix.net <http://www.edmatrix.net>

1 0