Mailman 3 October 2009 - Cplusplus-sig

RuntimeError on import from use of bases<>
by Craig S. Bosma Oct. 26, 2009

Oct. 26, 2009

Hi, I'm having a difficult time debugging an issue with boost.python bindings under OS X 10.6 -- the same bindings have been working fine under windows. On import of my module, I get an exception caught by python: > RuntimeError: unidentifiable C++ exception which I've traced down as far as the use of a bases<> entry in my class_ definition. If I remove the bases<> entry, it works, but of course removing all inheiritance from the module isn't an option. Furthermore, it seems that any (so far) use of bases<> in this module causes the problem, while in another module (wrapping another library) classes deriving from others via bases<> work normally. Any thoughts on what could case this type of failure? Craig

1 0

How to install extensions using bjam(?)
by Amos Anderson Oct. 26, 2009

Oct. 26, 2009

Hello -- I'm working on a project which is mixed C++/Python, and we use Boost functionality in our C++, and we've set up bjam to compile our project (on OSX). Everything seems to be working quite well. To use our libraries from python scripts right now, we have to run this first: #!/bin/bash trunk=$HOME/project/trunk/ export DYLD_LIBRARY_PATH=$DYLD_LIBRARY_PATH:$trunk/lib export PYTHONPATH=$trunk:$trunk/lib python $@ which is quite inconvenient. Now obviously I can skip this step by pasting those two export statements into my .profile, however, at some point we're going to start distributing this software and it would be nice if we didn't require the end user to do something like that. I can't believe that either of these 2 steps are how it's supposed to work... so how should I be doing it? Ideally, I want to set something up so that the end user only has to run 1 (one) command to build & install. Here are some things I've thought about: 1) Having bjam edit the user's .profile to set up these paths, possibly by calling a script I write. This doesn't seem like the right approach to me. 2) It looks like python's distutils is supposed to be able to help, but it seems like a waste of time to figure out how to get distutils to compile my code when I already have bjam pretty much figured out. First, I assume that distutils can put my libraries on an automatic DYLD_LIBRARY_PATH path that python knows about, because if it can't then there's no point. But assuming that's not a problem, I can't figure out how to tell distutils about Extensions that it shouldn't try to build... 3) bjam includes a way to install libraries into a folder I can name, but does it have a way to set environment variables? or automatically figure out the right places to put all the python files? If so, I can't find it... 4) ? thanks! Amos.

3 4

Help with converters ...
by Michele De Stefano Oct. 20, 2009

Oct. 20, 2009

Hello to everyone. I'm stuck with this problem. I'm trying to write a "from Python converter", from a tuple(tuple,tuple) type to C++. Clarifying more, the Python type is a tuple, containing 2 tuples. Now, when I try to implement the "convertible" static member function of the converter class, I write this (I've extracted only the interesting part): static void* convertible(PyObject* obj_ptr) { using namespace boost::python; handle<> hndl(borrowed(obj_ptr)); object in(hndl); if (!PyTuple_Check(in.ptr())) return NULL; extract<const tuple&> get_tup0(in[0]), get_tup1(in[1]); bool t1 = get_tup0.check(), t2 = get_tup1.check(); ...... the rest of the code is not necessary .... The problem is that even if I pass a tuple containing two tuples from Python, t1 and t2 become "false". Surely I'm not interfacing properly with Python, but I don't understand why. Can someone help me to understand ? Thank you in advance. Michele -- Michele De Stefano http://www.linkedin.com/in/micdestefano http://xoomer.virgilio.it/michele_de_stefano

2 2

Compile error when using exception_translation.hpp
by Ger van Diepen Oct. 13, 2009

Oct. 13, 2009

When compiling the following code using gcc-4.3.2 and boost-1.37, I get errors as listed at the end. It also fails for boost-1.40 and gcc-4.3.4. It is solved when patching translate_exception.hpp by adding # include <boost/type_traits/add_reference.hpp> Is it a Boost-Python problem or do I do something wrong? Of course, I can include that file in my own code, but that does not seem the correct solution to me. Thanks, Ger van Diepen #include <boost/translate_exception.hpp> #include <stdexcept> void translate_stdexcp (const std::exception& e) { PyErr_SetString(PyExc_RuntimeError, e.what()); } //# Note that the most general exception must be registered first. void register_convert_excp() { boost::python::register_exception_translator<std::exception> (&translate_stdexcp); } /home/jds/Work/svn-usg/release/include/boost/python/detail/translate_exception.hpp:34: error: expected nested-name-specifier before ‘add_reference’ /home/jds/Work/svn-usg/release/include/boost/python/detail/translate_exception.hpp:34: error: expected ‘;’ before ‘<’ token /home/jds/Work/svn-usg/release/include/boost/python/detail/translate_exception.hpp: In member function ‘bool boost::python::detail::translate_exception<ExceptionType, Translate>::operator()(const boost::python::detail::exception_handler&, const boost::function0<void>&, typename boost::call_traits<Translate>::param_type) const’: /home/jds/Work/svn-usg/release/include/boost/python/detail/translate_exception.hpp:56: error: expected type-specifier before ‘exception_cref’ /home/jds/Work/svn-usg/release/include/boost/python/detail/translate_exception.hpp:56: error: expected `)' before ‘e’ /home/jds/Work/svn-usg/release/include/boost/python/detail/translate_exception.hpp:56: error: expected `{' before ‘e’ /home/jds/Work/svn-usg/release/include/boost/python/detail/translate_exception.hpp:56: error: ‘e’ was not declared in this scope /home/jds/Work/svn-usg/release/include/boost/python/detail/translate_exception.hpp:56: error: expected `;' before ‘)’ token

2 1

object constructor + integer == fail
by Daniel Löb Oct. 13, 2009

Oct. 13, 2009

Hi! I seem to be unable to call the constructor of "object" with an integer parameter. Working example: #include <boost/python.hpp> #include <iostream> using namespace boost::python; using std::cout; using std::endl; int main() { object toast(3); cout << extract<int>(toast) << endl; } It compiles just fine, but dies with a segmentation fault on execution of the first line within main. Backtrace: 0x00007ffff7ac9ca3 in PyInt_FromLong () from /usr/lib/libpython2.5.so.1.0 #1 0x0000000000401860 in arg_to_python (this=0x7fffffffeb30, x=@0x7fffffffebdc) at /usr/include/boost/python/converter/builtin_converters.hpp:113 #2 0x0000000000401888 in boost::python::api::object_initializer_impl<false, false>::get<int> (x=@0x7fffffffebdc) at /usr/include/boost/python/object_core.hpp:374 #3 0x00000000004018cb in boost::python::api::object_base_initializer<int> (x=@0x7fffffffebdc) at /usr/include/boost/python/object_core.hpp:296 #4 0x00000000004018e7 in object<int> (this=0x7fffffffebd0, x=@0x7fffffffebdc) at /usr/include/boost/python/object_core.hpp:315 #5 0x0000000000401196 in main () at bp_bug.cpp:10 Doing the same as above with a string instead of an integer works fine. I'm running debian with libboost-python1.38.0 . Greetings, Daniel

2 2

Re: [C++-sig] Cplusplus-sig Digest, Vol 13, Issue 11
by Pim Schellart Oct. 13, 2009

Oct. 13, 2009

Dear Ravi, thanks for the link. I'll be sure to try that. Any chance this will be included (and documented) in boost python? Numpy is needed for just about any scientific software package using Python and it would be a real plus if boost Python had an easy to use interface to C++ for it. Kind regards, Pim Schellart 2009/10/13 <cplusplus-sig-request(a)python.org>: > Send Cplusplus-sig mailing list submissions to > cplusplus-sig(a)python.org > > To subscribe or unsubscribe via the World Wide Web, visit > http://mail.python.org/mailman/listinfo/cplusplus-sig > or, via email, send a message with subject or body 'help' to > cplusplus-sig-request(a)python.org > > You can reach the person managing the list at > cplusplus-sig-owner(a)python.org > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Cplusplus-sig digest..." > > > Today's Topics: > > 1. Re: [Py++] How to wrap default arg as enum with complicated > scope? (Roman Yakovenko) > 2. Re: [python] Function objects in place of member functions (Ravi) > 3. Re: Numpy ndarray as argument or return value using boost > python (Ravi) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Tue, 13 Oct 2009 00:39:04 +0200 > From: Roman Yakovenko <roman.yakovenko(a)gmail.com> > To: "Development of Python/C++ integration" <cplusplus-sig(a)python.org> > Subject: Re: [C++-sig] [Py++] How to wrap default arg as enum with > complicated scope? > Message-ID: > <7465b6170910121539i4b16cc8am5ab1d38b42fc474(a)mail.gmail.com> > Content-Type: text/plain; charset=ISO-8859-1 > > On Mon, Oct 12, 2009 at 10:51 PM, Christopher Bruns > <cmbruns(a)stanford.edu> wrote: >> Below is a simplified example of a tricky wrapping problem I have >> encountered. ?The example remains somewhat complicated, but further >> simplification results in a file that causes no trouble. >> I get a compile error when I try to build the boost python code >> generated by the following case: >> >> #### test_enum.h #### >> struct System { >> ? ?struct ProjectOptions { >> ? ? ? ?enum Option { >> ? ? ? ? ? ?All = 1 >> ? ? ? ?}; >> ? ? ? ?ProjectOptions(Option); >> ? ?}; >> ? ?void project(ProjectOptions = ProjectOptions::All); >> }; >> #### end test_enum.h #### >> >> The trouble comes from the default value ('ProjectOptions::All') for >> the project() method. ?Notice also the constructor for ProjectOptions >> from Options. ?That is required for reproducing the trouble here. >> >> The resulting boost.python code is as follows: >> ... >> #### test_wrap_enum2.cpp #### >> >> The symbol "All" in the line ", ( bp::arg("arg0")=All ) );" is out of >> scope here. ?If I change the line from >> >> ?, ( bp::arg("arg0")=All ) ); // compile failure >> >> to >> >> ?, ( bp::arg("arg0")=System::ProjectOptions::All ) ); // works >> >> it compiles fine. >> >> Is there any way to instruct pyplusplus to generate the default >> argument value for the project() method with the scope >> "System::ProjectOptions::All"? > > Yes: > > mb = module_builder_t(...) > project = mb.mem_fun( 'project' ) > project.arguments[0].default_value = "System::ProjectOptions::All" > >> Any enlightenment is much appreciated. > > Basically this is GCCXML limitations. The default argument could be > full-blown C++ expression. GCCXML doesn't handles\dumps them. It do > writes something for default arguments and Py++(pygccxml) try to > decode what it is. > > May be the following link could help you: > http://language-binding.net/pygccxml/upgrade_issues.html > http://language-binding.net/pyplusplus/documentation/functions/functions.ht… > > Also try google, this problem was discussed on this, pygccxml and > gccxml mailing lists. > > HTH > > -- > Roman Yakovenko > C++ Python language binding > http://www.language-binding.net/ > > > ------------------------------ > > Message: 2 > Date: Mon, 12 Oct 2009 21:04:46 -0400 > From: Ravi <lists_ravi(a)lavabit.com> > To: "Development of Python/C++ integration" <cplusplus-sig(a)python.org> > Subject: Re: [C++-sig] [python] Function objects in place of member > functions > Message-ID: <200910122104.46505.lists_ravi(a)lavabit.com> > Content-Type: Text/Plain; charset="iso-8859-1" > > On Monday 12 October 2009 08:47:22 troy d. straszheim wrote: >> >> boost::function<int(X*, int)> bf0(fobj); >> > >> > >> > Why do you need to use boost::function here? Shouldn't the type be >> > deduced automatically? >> > >> >> Note that the .def() of the various boost::function objects work without >> requiring the user to specialize get_signature, (since this modified >> get_signature metafunction knows about boost::function objects). > > [snip] > >> I think interoperability with boost::function is a Good Thing, but in >> the example above (and the previous), all it does is give def() a way to >> tell what the signature of the function object is at compile time, and >> it does so at a runtime cost. > > At least for my case, that extra virtual function call is a killer. > >> A specialization of get_signature works, >> but strikes me as kinda ugly (it should at least be moved out of >> 'detail' if it is a user customization point). How about something like >> >> class_<T>("T") >> .def("f1", f1, sig<void(T*, double)>) >> .def<void(T*, double)>("fi", f1) > > I'd rather have something along the lines of > .def< mpl::vector<void,T*,double> >("f1",f1) > by using function_types to compose the currect function signature. > Alternatively, one could have the sig<> template above be a def_visitor and > take the appropriate template arguments without the need to modify class_::def > at all. > >> >> I'm fairly new to the internals of boost.python, and only just now got >> >> this working... Do you see problems with this, specifically the >> >> conversion of get_signature from function to metafunction? >> > >> > >> > I don't see any problems with the conversion of get_signature to a >> > metafunction. Do compile times get any longer? >> >> I haven't checked this yet. I have other concerns re typechecking and >> automatic conversions... > > Care to elaborate? > > Regards, > Ravi > > > > > ------------------------------ > > Message: 3 > Date: Mon, 12 Oct 2009 21:11:44 -0400 > From: Ravi <lists_ravi(a)lavabit.com> > To: "Development of Python/C++ integration" <cplusplus-sig(a)python.org> > Subject: Re: [C++-sig] Numpy ndarray as argument or return value using > boost python > Message-ID: <200910122111.44171.lists_ravi(a)lavabit.com> > Content-Type: Text/Plain; charset="iso-8859-1" > > On Monday 12 October 2009 11:36:20 Pim Schellart wrote: >> 1. Is boost python still maintained or should I switch to another tool. >> 2. Is numpy supported, and if so, >> 3. can someone give me a basic example of how to use it. > > Please see > http://mail.python.org/pipermail/cplusplus-sig/2008-October/013825.html > which supports numpy arrays with and without copying, along with pass by > reference and pass by value. > > Regards, > Ravi > > > > ------------------------------ > > _______________________________________________ > Cplusplus-sig mailing list > Cplusplus-sig(a)python.org > http://mail.python.org/mailman/listinfo/cplusplus-sig > > End of Cplusplus-sig Digest, Vol 13, Issue 11 > ********************************************* >

1 0

Numpy ndarray as argument or return value using boost python
by Pim Schellart Oct. 13, 2009

Oct. 13, 2009

Hi Everyone, I also posted this to the boost users list. I tried to find a good example for this but after spending two days on google and one day getting pyublas to compile I decided to call for help. I want to give a numpy ndarray as an argument to a c++ method, calculate something and return the output as an ndarray to Python. I would prefer not to use another library (such as PyUblas) for this so I do not have to add another dependency to my program. Since the latest information on boost python is quite outdated (uses numeric) I would like to know the following. 1. Is boost python still maintained or should I switch to another tool. 2. Is numpy supported, and if so, 3. can someone give me a basic example of how to use it. Kind regards, Pim Schellart

4 3

[Py++] How to wrap default arg as enum with complicated scope?
by Christopher Bruns Oct. 12, 2009

Oct. 12, 2009

Below is a simplified example of a tricky wrapping problem I have encountered. The example remains somewhat complicated, but further simplification results in a file that causes no trouble. I get a compile error when I try to build the boost python code generated by the following case: #### test_enum.h #### struct System { struct ProjectOptions { enum Option { All = 1 }; ProjectOptions(Option); }; void project(ProjectOptions = ProjectOptions::All); }; #### end test_enum.h #### The trouble comes from the default value ('ProjectOptions::All') for the project() method. Notice also the constructor for ProjectOptions from Options. That is required for reproducing the trouble here. The resulting boost.python code is as follows: #### wrap_enum2.py #### from pyplusplus import module_builder mb = module_builder.module_builder_t(["test_enum2.h"] , gccxml_path = "gccxml.exe") mb.build_code_creator( module_name='test' ) mb.write_module( 'test_wrap_enum2.cpp' ) #### end wrap_enum2.py #### #### test_wrap_enum2.cpp #### // This file has been generated by Py++. #include "boost/python.hpp" #include "test_enum2.h" namespace bp = boost::python; BOOST_PYTHON_MODULE(test){ { //::System typedef bp::class_< System > System_exposer_t; System_exposer_t System_exposer = System_exposer_t( "System" ); bp::scope System_scope( System_exposer ); { //::System::ProjectOptions typedef bp::class_< System::ProjectOptions > ProjectOptions_exposer_t; ProjectOptions_exposer_t ProjectOptions_exposer = ProjectOptions_exposer_t( "ProjectOptions", bp::init< System::ProjectOptions::Option >(( bp::arg("arg0") )) ); bp::scope ProjectOptions_scope( ProjectOptions_exposer ); bp::enum_< System::ProjectOptions::Option>("Option") .value("All", System::ProjectOptions::All) .export_values() ; bp::implicitly_convertible< System::ProjectOptions::Option, System::ProjectOptions >(); } { //::System::project typedef void ( ::System::*project_function_type )( ::System::ProjectOptions ) ; System_exposer.def( "project" , project_function_type( &::System::project ) , ( bp::arg("arg0")=All ) ); } } } #### test_wrap_enum2.cpp #### The symbol "All" in the line ", ( bp::arg("arg0")=All ) );" is out of scope here. If I change the line from , ( bp::arg("arg0")=All ) ); // compile failure to , ( bp::arg("arg0")=System::ProjectOptions::All ) ); // works it compiles fine. Is there any way to instruct pyplusplus to generate the default argument value for the project() method with the scope "System::ProjectOptions::All"? Any enlightenment is much appreciated. --Chris Bruns

2 1

Returning pointer to abstract class from pure virtual class method
by Roman Hwang Oct. 12, 2009

Oct. 12, 2009

Hello. I'm new to Python, but really want to use it with my C++ code. I faced a problem while returning pointer to abstract class from pure virtual class method. I tried to search the web, but found nothing. struct ICoreWrapper : ICore, wrapper<ICore> { virtual IRegistry* CreateRegistry() { return this->get_override("CreateRegistry")(); } }; struct IRegistryWrapper : IRegistry, wrapper<IRegistry> { virtual void DoSomestuff() { return this->get_override("DoSomestuff")(); } }; ICore* CreateCoreW() { return new CCore; } BOOST_PYTHON_MODULE_INIT(CorePy) { class_<IRegistryWrapper, boost::noncopyable>("IRegistry") .def("DoSomestuff", pure_virtual(&IRegistry::DoSomestuff)) ; class_<ICoreWrapper, boost::noncopyable>("ICore") .def("CreateRegistry", pure_virtual(&ICore::CreateRegistry), return_value_policy<reference_existing_object>()) ; def("CreateCore", CreateCoreW, return_value_policy<reference_existing_object>()); } The Python code looks like this: >>> import CorePy >>> c = CorePy.CreateCore() >>> b = c.CreateRegistry() I get Debug Error Run-Time Check Failure #0 - The value of ESP was not properly saved across a function call. This is usually a result of calling a function declared with one calling convention with a function pointer declared with a different calling convention. Please tell me what I do wrong. Or whether this code is supported or not. Thanks Roman Hwang.

1 0

Boost.python: extracting an array
by Randolph Fritz Oct. 12, 2009

Oct. 12, 2009

Is there any way to extract a vector of floats from a python array? In Python, can I write something like: from array import array from bpclass import foo vec = array('f', (1,2,3)) foo ([vec]) And, in C++: void bpclass::foo (list v) { ... extract< {magic} >(v[0]) ... Or would it be better to replace the array with a list? -- Randolph Fritz design machine group, architecture department, university of washington rfritz(a)u.washington.edu -or- rfritz333(a)gmail.com

1 0