Mailman 3 August 2006 - Cplusplus-sig

Re: [C++-sig] RuntimeError: unidentifiable C++ exception
by Qun Cao Aug. 31, 2006

Aug. 31, 2006

> > >* Hi Roman, > *>* What do you mean by "the code generated by Py++ just works"? Do you mean > *>* you created a class > *>* MyClass { > *>* public: > *>* MyClass(char const*) ; > *>* } > *>* and you had successfully initialized it in Python? I tried a similar > *>* example class and had no problem with it. > * > That what you described :-). > > >* The real problem I had was with this specific class NiString in Gamebryo > *>* game engine, is it possible that something > *>* going on inside Gamebryo might cause the problem? Or the game engine is > *>* irrelevant to this problem? > * > I think that NiString is relevant. Take a look on this FAQ: > http://www.boost.org/libs/python/doc/v2/faq.html#custom_string > > Hi Roman, Thanks for you suggestions. I looked into the custom_string code, it is a nice example but different with my problem. I don't want python automatically convert my NiFixedString to python string, all I want is a way to use the constructor of NiFixedString class to create a python object. a = NiFixedString("test") Theoretically this is no different than wrapping any other class with a constructor that takes a string/char const*, and this code should do it, right? BOOST_PYTHON_MODULE(NiStringTest) { class_<NiFixedString,boost::noncopyable>("pNiFixedString", init<char const*>()); } Btw, NiFixedString encapsulates a memory efficient implementation of reference-counted ASCII strings, here is the header file, is there any thing that potentially cause the exception? Also, from my understanding, PY++ simplifies the process of generating the wrapper code, could adopting PY++ possibly help with this problem? #ifndef NIFIXEDSTRINGTABLE_H #define NIFIXEDSTRINGTABLE_H #include "NiRTLib.h" #include "NiGlobalStringTable.h" #include "NiPath.h" class NIMAIN_ENTRY NiFixedString : public NiMemObject { public: NiFixedString(); NiFixedString(const char* pcString); NiFixedString(const NiFixedString& kString); ~NiFixedString(); operator const char*() const; bool Exists() const; NiFixedString& operator=(const NiFixedString& kString); NiFixedString& operator=(const char* pcString); size_t GetLength() const; unsigned int GetRefCount() const; bool Equals(const char* pcStr) const; bool EqualsNoCase(const char* pcStr) const; bool Contains(const char* pcStr) const; bool ContainsNoCase(const char* pcStr) const; friend bool operator==(const NiFixedString& s1, const NiFixedString& s2); friend bool operator!=(const NiFixedString& s1, const NiFixedString& s2); friend bool operator==(const NiFixedString& s1, const char* s2); friend bool operator!=(const NiFixedString& s1, const char* s2); friend bool operator==(const char* s1, const NiFixedString& s2); friend bool operator!=(const char* s1, const NiFixedString& s2); // Begin Emergent internal use only // End Emergent internal use only protected: NiGlobalStringTable::GlobalStringHandle m_kHandle; }; NIMAIN_ENTRY void NiStandardizeFilePath(NiFixedString& kString); #include "NiFixedString.inl" #endif Excerpt from NifixedString.inl inline NiFixedString::NiFixedString(const char* pcString) { if (pcString == NULL) { m_kHandle = NiGlobalStringTable::NULL_STRING; } else { m_kHandle = NiGlobalStringTable::AddString(pcString); } } >* And still, what the "unidentifiable C++ exception" is really saying? > * > Boost.Python has next code: > > try{ > run you C++ code > } > catch( registered exceptions ){ > convert them to Python exception > } > catch( ... ){ > throw Python RuntimeError( "unidentifiable C++ exception" ) > } > > So or you forgot to register your exception or ... something really > bad happened. > >

2 3

Describing how boost.python works
by Allen Bierbaum Aug. 31, 2006

Aug. 31, 2006

Hello all, I don't know about the rest of you, but I have been using boost.python for a number of years now and there is still a lot about how it works behind the scenes that seems like magic to me. Normally this isn't a bad thing, but when trying to track down bugs it becomes a problem because I just don't know what the system is trying to do. Earlier this year I became interested enough to spend several evenings looking through old mailing list questions and through the code to try to start documenting it for myself. I have put up the results of this effort on a wiki at: https://realityforge.vrsource.org/view/PyOpenSG/HowBoostPythonWorks There is still much to be documented (especially the conversion of data across the language barrier), but it is a start and hopefully it will be helpful to someone. Does anyone know of any more complete documentation of how boost.python works? Are there any other people willing to collaborate on documenting the inner workings of the system? -Allen

1 0

Trouble choosing boost python policy
by Eduardo Picado Aug. 31, 2006

Aug. 31, 2006

Hi, I'm wrapping the following function: /** * System retrieving. * \throw std::bad_alloc memory allocation error. * \return a reference to the system associated to the calling process. */ static TASRSystem & getASRSystem() throw (std::bad_alloc); Which policy should I choose? I read the individual doc for all of them, but it's a bit hard to get it. I imagine it might be return_internal_reference but it is not working… >>> system = TASRSystem.getASRSystem() Traceback (most recent call last): File "<stdin>", line 1, in ? IndexError: boost::python::with_custodian_and_ward_postcall: argument index out of range Thanks, Eduardo -- Eduardo PICADO MSN eduardo.picado(a)gmail.com portable +33 6 32 47 10 10 (NEW) fix +33 1 45 03 52 54 |̲̅<̲̅Θ̲̅>̲̅|

2 1

Re: [C++-sig] [Python] null pointer access violation when calling Python function
by David Abrahams Aug. 31, 2006

Aug. 31, 2006

Tanguy Fautre <tanguy.fautre(a)spaceapplications.com> writes: >>And please bring >> your Boost.Python questions to the C++-sig >> <http://boost.org/more/mailing_lists.htm#cplussig>. > > > After doing a minimal example reproducing this error, I think I found > the problem. > > According to the C++ Programming Language (3rd ed), Section 14.7: "It is > implementation-defined whether destructors are invoked when a program is > terminated because of an uncaught exception.". Yes. > That, I did not know. I thought destructor were always called. > > GCC 4.x and Visual C++ 2005 are indeed calling std::terminate without > calling the destructors. Except... Except when debugging with VC++ 2005. > When debugging, you have the choice to continue after a uncaught > exception is thrown as if nothing happened. This effectively leads to a > segfault as described as above. I think changing the semantics of "throw x" from "unwind the stack and maybe terminate" to "continue as though nothing has happened" seems like it could break any program. > So it's not a problem with my application nor Boost.Python. Just my > misunderstanding of uncaught exceptions and of Visual C++. > > Sorry I bothered you with this false alert. 'S ok. -- Dave Abrahams Boost Consulting www.boost-consulting.com

1 0

Py++ issues, plus switching from Swig
by wz Aug. 31, 2006

Aug. 31, 2006

Those links were indeed helpful, especially the sizeof() method once again shows that beauty is in simplicity. Still, my concerns are not completely vanished. 1) If I understood it right, in order to make my instanciated vector<float> named vector_Float in python, do I have to use this construct: typedef vector<float> vector_Float; sizeof(vector_Float); ? Is there a better way that would not let me write out all 1000s of templates possibilities from over 250 included headers? A simple mb.collect_templates('vector', ['float', 'bool', 'Point3', 'Quaternion']) would be a godsend. 2) How to deal with templated functions? As there are at least one DynamicCast<T> and one StaticCast<T> for each class, how can I automate their use in Py++? As said, I cannot manually make C++ instantiations of the classes. They change too frequently, and I could not catch up to their rate of change. Thanks again, N. :)

2 1

boost python doc problem? (return_self)
by Neal Becker Aug. 31, 2006

Aug. 31, 2006

The example in ref man for return_self, shows: class_<widget>("Widget") .def("sensitive", &Widget::get_sensitive) .def("sensitive", &Widget::set_sensitive, return_self<>()) ; But Widget::set_sensitive is void, it doesn't return anything. What am I missing here?

2 1

Py++ 0.8.1
by Roman Yakovenko Aug. 31, 2006

Aug. 31, 2006

I'm glad to announce the new version of Py++. Download page: http://language-binding.net/pyplusplus/download.html What is it? Py++ is an object-oriented framework for creating a code generator for Boost.Python library. Project home page: http://language-binding.net/pyplusplus/pyplusplus.html Changes: 1. Support for Boost.Python indexing suite version 2 was implemented. Indexing suite 2, provides support for almost all STD containers. 2. Py++ generates "stable" code. If header files were not changed, Py++ will not change any file. 3. Support for huge classes was added. Py++ is able to split registration code for the class to multiple cpp files. 4. User code could be added almost anywhere, without use of low level API. 5. Py++ will provide user with feedback: warnings about exported declarations. 6. Py++ now supports generation of documentation strings. 7. A lot of bugs were fixed. 8. Documentation was written. -- Roman Yakovenko C++ Python language binding http://www.language-binding.net/

2 1

Whether and how to use Py++ a template-loaded case
by Nicolas Tessore Aug. 30, 2006

Aug. 30, 2006

Hello, I hope someone can help me out: I am relatively new to Python, and not too experienced in C++. My situation is the following: I am trying to add Boost.Python bindings to a library consisting of about a hundred includes, weighting 60Megs. It is heavily relying on inheritance, in fact all classes derive from one base. Now the tricky bit: The lib uses templates a _lot_. The small part that is currently included in a Swig wrapper already contains 800 specific template conversions. What I now want to do is switch to Boost.Python for a change. Swig creates a huge number of problems when using the lib, and I trust Boost to perform better. Still there is one issue: Due to the number of classes, templates and the frequent changes of the codebase, I cannot generate the Boost.Python wrapper manually. I tried using Pyste, but it is still too much work to add all classes in an interface file. The only possibility is automation, as presented in Py++. The reason for posting is that I cannot find the documentation explaining how to do templates in Py++. I can (and do) use typedefs + declaration for templated types, but I need support for template functions as well, since I need to access a lot of Dynamic and Static casts, in fact, each single class needs it. Could anyone give me a hint in the right direction, or at least tell me that it is impossible in general, so I can hack the Swig interface to at least stop segfault'ing? Thanks in advance, N.

1 0

RuntimeError: unidentifiable C++ exception
by Qun Cao Aug. 30, 2006

Aug. 30, 2006

>* Hi Everyone, *>* *>* I created a simple class wrapper like this *>* *>* BOOST_PYTHON_MODULE(PythonTest) { *>* class_<MyClass,boost::noncopyable>("pMyClass", init<char const*>()); *>* } *>* *>* When I try to initialize pMyClass in python, I got this error: *>* *>* Traceback (most recent call last): *>* File "<stdin>", line 1, in ? *>* RuntimeError: unidentifiable C++ exception * The code generated by Py++ just works, may be you missed something? You can try to use make_consructor function: http://boost.org/libs/python/doc/v2/make_function.html#make_constructor-spec Hi Roman, What do you mean by "the code generated by Py++ just works"? Do you mean you created a class MyClass { public: MyClass(char const*) ; } and you had successfully initialized it in Python? I tried a similar example class and had no problem with it. The real problem I had was with this specific class NiString in Gamebryo game engine, is it possible that something going on inside Gamebryo might cause the problem? Or the game engine is irrelevant to this problem? And still, what the "unidentifiable C++ exception" is really saying? Boost-Python called the C++ constructor code and an exception was thrown? I am taking a look at Py++ and make_constructor, Thanks so much for your help! Qun

2 1

Py++ issues, plus switching from Swig
by Nicolas Tessore Aug. 30, 2006

Aug. 30, 2006

Hello, I hope someone can help me out: I am relatively new to Python, and not too experienced in C++. My situation is the following: I am trying to add Boost.Python bindings to a library consisting of about a hundred includes, weighting 60Megs. It is heavily relying on inheritance, in fact all classes derive from one base. Now the tricky bit: The lib uses templates a _lot_. The small part that is currently included in a Swig wrapper already contains 800 specific template conversions. What I now want to do is switch to Boost.Python for a change. Swig creates a huge number of problems when using the lib, and I trust Boost to perform better. Still there is one issue: Due to the number of classes, templates and the frequent changes of the codebase, I cannot generate the Boost.Python wrapper manually. I tried using Pyste, but it is still too much work to add all classes in an interface file. The only possibility is automation, as presented in Py++. The reason for posting is that I cannot find the documentation explaining how to do templates in Py++. I can (and do) use typedefs + declaration for templated types, but I need support for template functions as well, since I need to access a lot of Dynamic and Static casts, in fact, each single class needs it. Could anyone give me a hint in the right direction, or at least tell me that it is impossible in general, so I can hack the Swig interface to at least stop segfault'ing? Thanks in advance, N. -- "Feel free" – 10 GB Mailbox, 100 FreeSMS/Monat ... Jetzt GMX TopMail testen: http://www.gmx.net/de/go/topmail

2 1