Mailman 3 June 2007 - Cplusplus-sig

Using boost.python to get STL containers of PyObect's.
by Eric Mahurin June 26, 2007

June 26, 2007

Hi, I'm new to the list and coming from using SWIG w/ python. I haven't used boost.python, but am considering it because of frustration with SWIG. With SWIG, I've been able to make changes to some of the existing SWIG/STL infrastructure to get to this point: * STL containers of a wrapped PyObject* (handles reference counting) * Support of using a callable PyObject* as a BinaryPredicate and other template function concepts in STL. This allows using a python lamda/function/method as the comparison function for various methods/constructors. * python read/write access to STL iterators * wrap any STL algorithm to be used with the above wrapped STL iterators. I should have made it work with any python object that implemented the python-STL iterator i/f, but for now it has to be one of these swig-wrapped STL iterators. Here are the problems: * various STL methods are missing, especially some of the overloaded methods. * STL big-O performance is not alway preserved. I think I've fixed the major issues (I've seen O(n**3) when it should be O(n)), but I'm sure there are others. * hacky and incomplete iterator API I'm getting tired of wading through the multiple layers and confusion that SWIG puts in to finish this. I think what is most bothersome is the way type conversion and overloading is handled. I'd rather see SWIG only try to distinguish between overloaded functions/methods by number of arguments. You'd be forced to rename otherwise. IMHO, overloading by type doesn't mix with a dynamically typed language (i.e. duck-typed). Has anybody done some similar with boost.python? Eric

2 1

problems building Boost.Python with cygwin
by Knox, Matt June 22, 2007

June 22, 2007

Hello all, Apologies in advance if this is not the correct forum for this... but I am trying to install the Boost.Python library (and then compile C++ python extensions after that) using cygwin and am not having much luck. I know I have the main part of boost working because I was able to compile and run the basic example for testing the boost configuration. Then to install Boost.Python I do the following: $ ./configure -with-libraries=python This generates the following output: Building Boost.Jam with toolset gcc... tools/jam/src/bin.cygwinx86/bjam Detecting Python version... 2.5 Detecting Python root... c:\Python25 Unicode/ICU support for Boost.Regex?... not found. Backing up existing Boost.Build configuration in user-config.jam.2 Generating Boost.Build configuration in user-config.jam... Generating Makefile... I then type "make install", which generates the following output: ./tools/jam/src/bin.cygwinx86/bjam --user-config=user-config.jam --prefix=/usr/local --exec-prefix= /usr/local --libdir=/usr/local/lib --includedir=/usr/local/include --with-python install WARNING: No python installation configured and autoconfiguration failed. See http://www.boost.org/libs/python/doc/building.html for configuration instructions or pass --without-python to suppress this message and silently skip all Boost.Python targets warning: Python location is not configured warning: the Boost.Python library won't be built warning: Unable to construct ./install-unversioned warning: Unable to construct ./install-unversioned warning: Unable to construct ./install-unversioned warning: Unable to construct ./install-unversioned warning: Unable to construct ./install-unversioned warning: Unable to construct ./install-unversioned warning: Unable to construct ./install-unversioned warning: Unable to construct ./install-unversioned ...found 7812 targets... Those warnings look rather problematic. And indeed, when I try to compile the basic example (/boost_1_34_0/libs/python/example/tutorial) by typing "bjam target-os=cygwin toolset=gcc", I get the following errors: Jamroot:17: in modules.load rule python-extension unknown in module Jamfile</cygdrive/p/c_libs/boost_1_34_0/libs/python/example/ tutorial>. /cygdrive/p/c_libs/boost_1_34_0/tools/build/v2/build/project.jam:312: in load-jamfile /cygdrive/p/c_libs/boost_1_34_0/tools/build/v2/build/project.jam:68: in load /cygdrive/p/c_libs/boost_1_34_0/tools/build/v2/build/project.jam:170: in project.find /cygdrive/p/c_libs/boost_1_34_0/tools/build/v2/build-system.jam:237: in load /cygdrive/p/c_libs/boost_1_34_0/libs/python/example/../../../tools/build /v2/kernel/modules.jam:261: in import /cygdrive/p/c_libs/boost_1_34_0/libs/python/example/../../../tools/build /v2/kernel/bootstrap.jam:132 : in boost-build /cygdrive/p/c_libs/boost_1_34_0/libs/python/example/boost-build.jam:7: in module scope Any ideas? Am I missing something obvious? Any help would be greatly appreciated. Thanks, - Matt - ******************** NOTICE OF CONFIDENTIALITY This communication including any information transmitted with it is intended only for the use of the addressees and is confidential. If you are not an intended recipient or responsible for delivering the message to an intended recipient, any review, disclosure, conversion to hard copy, dissemination, reproduction or other use of any part of this communication is strictly prohibited, as is the taking or omitting of any action in reliance upon this communication. If you receive this communication in error or without authorization please notify us immediately by return e-mail or otherwise and permanently delete the entire communication from any computer, disk drive, or other storage medium. If the above disclaimer is not properly readable, it can be found at www.td.com/legal AVERTISSEMENT DE CONFIDENTIALITE Ce courriel, ainsi que tout renseignement ci-inclus, destiné uniquement aux destinataires susmentionnés, est confidentiel. Si vous nêtes pas le destinataire prévu ou un agent responsable de la livraison de ce courriel, tout examen, divulgation, copie, impression, reproduction, distribution, ou autre utilisation dune partie de ce courriel est strictement interdit de même que toute intervention ou abstraction à cet égard. Si vous avez reçu ce message par erreur ou sans autorisation, veuillez en aviser immédiatement lexpéditeur par retour de courriel ou par un autre moyen et supprimer immédiatement cette communication entière de tout système électronique. Si l'avis de non-responsabilité ci-dessus n'est pas lisible, vous pouvez le consulter à www.td.com/francais/legale

1 0

boost::python, PyObject* members, and cyclic references
by Jim Bosch June 21, 2007

June 21, 2007

I am trying to wrap a class in which I'd like to expose some of its data as a numpy array, so in python it works like this: >>> o = array_holder() >>> a = o.array >>> a[4:6] = 3.14 Rather than constructing the array every time, I'd like to store the PyObject* (or something equivalent) in my array_holder wrapper, but I need to tie the lifetimes of 'o' and 'a' together, so nothing bad happens if someone does 'del o' while 'a' still exists. The numpy array I'm constructing holds a reference to the PyObject that owns its data (which I assume I can get from 'object(this).ptr()'), so I'm not worried about that end. And I assume something like 'return_internal_reference' is what I'm looking for at the other end, but I'm not quite sure how to work out the details (do I return a 'PyObject*', a 'handle', or an 'object'?) Finally, there's the question of whether either of these will ever get deleted at all - do boost::python wrapped objects participate in cyclic gc? If not, is there any way I can manually add it for this class (by suppling traverse and clear methods using the C API, for instance)? Thanks! Jim Bosch

2 1

Py++: absolute or relative PATH for parsed files
by Adrien Saladin June 20, 2007

June 20, 2007

Hello, I am switching from Pyste to Py++ for my project. Thanks for this great tool ! (and for boost python of course). I still have some problem with absolute and relative file paths: I have a file called "all.h" which only contains #include directives for the other .h files. In the Py++ driver file, when I use this first line, the bindings are generated: mb = module_builder.module_builder_t(files=[r"/nfs/to/my/home/Src/Tests/Python/Py++/all.h"]) As this way is not portable because of the absolte path, I just cut this line to mb = module_builder.module_builder_t(files=[r"all.h"]) and now the generated module is empty in my example directory. I attached files needed to reproduce the problem. What am I doing wrong ? Sorry if this question is trivial. I'm using Py++ 0.9 (installed from sources) pygccxml-0.9 and gccxml 0.7.0+cvs20060311-2 (debian Etch) Best Regards, Adrien Saladin

2 2

Share C++ Object Extended in Python?
by David Brownell June 19, 2007

June 19, 2007

I have an existing C++ definition that resembles the code below, and would like to extend the Events interface using Python. However, I get errors in Python when I attempt to create a Python object that is shared between Python and C++. struct Events { virtual void Foo(char const *tcString) = 0; }; struct Object { void Init(Events &events) { _pEvents = &events; } void DoFoo() { _pEvents->Foo("Fire foo!"); } Events * _pEvents; }; When using the following boost python code generated with Py++ things work well in some cases, but not all. struct Events_wrapper : Events, bp::wrapper< Events > { Events_wrapper() : Events() , bp::wrapper< Events >(){ // null constructor } virtual void Foo( char const * tcString ){ bp::override func_Foo = this->get_override( "Foo" ); func_Foo( tcString ); } }; BOOST_PYTHON_MODULE(tester){ bp::class_< Events_wrapper, boost::noncopyable >( "Events" ) .def( "Foo" , bp::pure_virtual( &::Events::Foo ) , ( bp::arg("tcString") ) ); bp::class_< Object >( "Object" ) .def( "DoFoo" , &::Object::DoFoo ) .def( "Init" , &::Object::Init , ( bp::arg("events") ) ); } This following Python code works fine: import tester class Events(tester.Events): def Foo(self s): print s e = Events() o = tester.Object() o.Init(e) o.DoFoo() But I get the error... Traceback (most recent call last): File "<interactive input>", line 1, in ? ArgumentError: Python argument types in Object.Init(Object, Events) did not match C++ signature: Init(struct Object {lvalue}, struct Events {lvalue} events) ...with the code: import tester class Events(tester.Events): def __init__(self): self.i = 0 def Foo(self s): print s e = Events() o = tester.Object() o.Init(e) o.DoFoo() It appears that I am not doing something right with the binding code, but so far I haven't been able to find the right combination of boost::python argument decorators to get the job done. It seems like I could use an auto_ptr if I wanted Object to take ownership of the event object, but I want both the Python and C++ code to continue to modify the same instance of Event. Is there some way I can do this? Thanks for your help! David Brownell

2 2

Py++ and 'using' declarations
by Pertti Kellomäki June 19, 2007

June 19, 2007

Is there a way in Py++ to deal with overloaded member functions introduced by 'using' declarations? For example, for the following declarations, Py++ creates the code below. ----------------------------------------------------- class B { public: void f(); }; class D : public B { public: void f(int i); using B::f; }; ----------------------------------------------------- // This file has been generated by Py++. #include "boost/python.hpp" #include "killme.hh" namespace bp = boost::python; BOOST_PYTHON_MODULE(killme){ bp::class_< B >( "B" ) .def("f", &::B::f ); bp::class_< D, bp::bases< B > >( "D" ) .def("f", &::D::f, ( bp::arg("i") ) ); } ----------------------------------------------------- Compiling this with g++ results in an "unresolved overloaded function type" error, because g++ cannot choose between D::f() and D::f(int). -- Pertti

2 2

r-value converter
by Piotr Jaroszyński June 17, 2007

June 17, 2007

Hello, for one of the classes I had to write PythonFoo equivalent and hence I need a r-value converter from PythonFoo to Foo as implicit conversion is not possible. This what I have (more or less copy-n-paste from Roman's and boost's docs :) struct PackageDepSpecFromPython { PackageDepSpecFromPython() { bp::converter::registry::push_back(&convertible, &construct, boost::python::type_id<PackageDepSpec>()); } static void * convertible(PyObject * obj_ptr) { return obj_ptr; } static void construct(PyObject * obj_ptr, bp::converter::rvalue_from_python_stage1_data * data) { typedef bp::converter::rvalue_from_python_storage<PythonPackageDepSpec> Storage; void * storage = reinterpret_cast<Storage *>(data)->storage.bytes; new (storage) PackageDepSpec("cat/123", pds_pm_permissive); [1] data->convertible = storage; } }; And a few questions: 1) First wrt [1], I would like to create the PackageDepSpec with properties taken from PythonPackageDepSpec equivalent - how can I get that C++ object from PyObject pointer? 2) Some of the functions taking PackageDepSpec want shared_ptr<PackageDepSpec> - how can I make a converter to that type? 3) Should I do any extra checks in convertible for such a custom type? -- Best Regards, Piotr Jaroszyński

1 2

Exception translation
by Albert Strasheim June 16, 2007

June 16, 2007

Hello all I want to translate some of my exceptions from C++ to Python. I have registered an exception translator and it works rather well. However, I would like my translated exceptions to have their own type in Python, so I don't simply want to do: PyErr_SetString(PyExc_UserWarning, ex.what()); in my exception translator. Instead, I want my exception translator to take the exception, clone it (luckily I have an exception base class that allows me to do this), construct an instance of its wrapper type around it and use PyErr_SetObject instead. Something along these lines has been discussed previously: http://mail.python.org/pipermail/c++-sig/2006-April/010618.html Later in thread, Dave Abrahams said: Officially, it [Python] does require that exceptions be derived from PyExc_Exception (I don't remember where that's documented), but unofficially, you can throw anything :) I assume this policy exists so that users can write try/except blocks that catch Exception and have it catch any type derived from Exception. Now my question: is there a way to derive the Boost.Python wrapper type of my C++ exception class from PyExc_Exception using Boost.Python? I want to do this so that my exceptions can be caught in scenario described above. Even better, is there a way to derive an arbitrary Boost.Python wrapper type of any C++ class from an arbitrary Python type? This might turn out to be very handy for integrating C++ matrix libraries with NumPy. Thanks for any help. Cheers, Albert P.S. In thread, Roman Yakovenko mentioned that he might add support for this kind of thing to pyplusplus. Did something happen in this regard?

3 4

Re: [C++-sig] len function?
by Ralf W. Grosse-Kunstleve June 15, 2007

June 15, 2007

> > Thanks, Alex, this is what I ended up doing. Why is the len() function > > in the api_placeholder header? Shouldn't that be a standard > > boost.python function? > > I agree with you, but for an official response we'd have to ask David > Abrahams. My guess is that his response would be, "Patches are welcome." In the Boost CVS head and the 1.34.0 release, len() is defined in boost/python/object.hpp (ever since Jan or Feb 2006). api_placeholder.hpp is still there but only for backward compatibility. I'll remove it from the head in a year or two. Ralf ____________________________________________________________________________________ Bored stiff? Loosen up... Download and play hundreds of games for free on Yahoo! Games. http://games.yahoo.com/games/front

2 1

bgl Python on Mac OS X
by Peter Saffrey June 14, 2007

June 14, 2007

I'm trying to build this. I've downloaded the boost libraries (version 1.34) and run configure, make and sudo make install without errors. Then I've downloaded bgl-python (0.9) and, as instructed, found the bjam executable from the boost hierarchy, copied it somewhere convenient and run it in the bgl-python directory. That gave me this error: Unable to load Boost.Build: could not find "boost-build.jam" --------------------------------------------------------------- Attempted search from /Users/pzs/incoming/bgl-python-0.9 up to the root and in these directories from BOOST_BUILD_PATH and BOOST_ROOT: /usr/ share/boost-build. Please consult the documentation at 'http://www.boost.org'. So I set $BOOST_ROOT to be the root of the boost 1.34 directory and tried bjam again. This time I get: error: Could not find parent for project at '.' error: Did not find Jamfile or project-root.jam in any parent directory. I've also tried the binary, but that doesn't work either, probably because I'm running Python 2.5 and the binary is for 2.3. Any ideas? Peter

2 1