Mailman 3 July 2003 - Cplusplus-sig

RE: [C++-sig] Re: def_readonly and const array members
by Jeff Brewer July 27, 2003

July 27, 2003

Sorry about the error message difference - nothing funny is going on. I tested with an 8 character array first and then a 16. I included the error message from the 8 try. I'll try it with an int array and post the results. Jeff Brewer -----Original Message----- From: Ralf W. Grosse-Kunstleve [mailto:rwgk@yahoo.com] Posted At: Friday, July 25, 2003 9:00 AM Posted To: c++-sig Conversation: [C++-sig] Re: def_readonly and const array members Subject: Re: [C++-sig] Re: def_readonly and const array members --- David Abrahams <dave(a)boost-consulting.com> wrote: > "Jeff Brewer" <jeff(a)brewer.com> writes: > > > I am using Boost 1.30. Sorry I forgot to mention that. > > > > def_readonly seems to still in 1.30 and its implementation is: > > template <class D, class B> > > self& def_readonly(char const* name, D B::*pm_) > > { > > D T::*pm = pm_; > > this->add_property(name, make_getter(pm)); > > return *this; > > } > > Is it deprecated though? > > Nope. I don't have time to figure out why this isn't working for you > right now, but... If David cannot figure it out, maybe it is time to admit that I am lost, too. I am doing all the time what you are trying to do, under many platforms including Visual C++ 7.1. However, I've never tried it with char arrays. What happens if you change your small test to work with, say, int instead of char? I am not sure it will help, but here is my "to_tuple" converter: template <typename ContainerType> struct to_tuple { static PyObject* convert(ContainerType const& a) { boost::python::list result; for(std::size_t i=0;i<a.size();i++) { result.append(boost::python::object(a[i])); } return boost::python::incref(boost::python::tuple(result).ptr()); } }; Your's looks good, too, but maybe there is a subtlety that VC chokes on? I also noticed that your test involves char array[16], but you quote an error message related to char array[8]? Where does that factor of two come from? OK, enough shots in the blue for now. Hope this is useful somehow. Ralf __________________________________ Do you Yahoo!? Yahoo! SiteBuilder - Free, easy-to-use web site design software http://sitebuilder.yahoo.com _______________________________________________ C++-sig mailing list C++-sig(a)python.org http://mail.python.org/mailman/listinfo/c++-sig

2 1

introspecting functions
by Daniel Holth July 27, 2003

July 27, 2003

C++-siggers, I have looked into how pydoc works. It doesn't look too bad. Pydoc does a large part of its work using the inspect module to look at code objects. 'Real' Python functions have a func_code property. Example: >>> urlparse.urlsplit.func_code.co_varnames ('url', 'scheme', 'allow_fragments', 'i', 'cached', 'c', 'netloc', 'fragment', 'tuple', 'key', 'query') And the code for urlsplit is: def urlsplit(url, scheme='', allow_fragments=1): key = url, scheme, allow_fragments cached = _parse_cache.get(key, None) if cached: return cached if len(_parse_cache) >= MAX_CACHE_SIZE: # avoid runaway growth clear_cache() netloc = query = fragment = '' i = url.find(':') if i > 0: if url[:i] == 'http': # optimize the common case scheme = url[:i].lower() url = url[i+1:] if url[:2] == '//': i = url.find('/', 2) ...... and on it goes. Anyway, the first expression gives me all the variable names in the function. The first three .co_varnames appear to be the function argument names (.co_argcount). You can also retrieve the filename the function was compiled from and the bytecode for the function. I don't know what the consequences of using a fake func_code are and I would have played with it more but I don't know how to best add a func_code to my boost::python classes, but it's an interesting idea. You could probably figure out how to implement the equivalent of def foo(bar=1, baz=2) for the C++ for example. Whoah! Pydoc actually parses the bytecodes for classes in order to document what the default variable values are (not what the variable names are though). It mightn't be outrageous to compile lots of little def wrappedfunction(a, b=None): pass in order to make some bytecode though, or even replace : pass with : call_boost_function(). - Daniel Holth

2 1

Wrapping Just an Abstract Base
by Thomas Muldowney July 26, 2003

July 26, 2003

I'm having a mental breakdown on this one. I have a factory returning a pointer to an abstract base class that I've wrapped. However, I can't logically wrap the actual derived class because it's generated in a plugin. So, when I go to make a call on one of the abstract functions it fails with an AttributeError exception. Is there anyway to handle this in a clean manner that keeps the API near the C++ API? I've played around with writing wrapper functions (taking the base class pointer) that are defined on the class using the same name as the virtual function. Sadly, from my mailing list searches on this subject it seems this is a bad idea and can possibly lead to lookup loops? It does seem to run properly, though. Would love some feedback or a kick in the head if I'm missing something in the docs and sites. Thanks. --temas

2 4

building boost.python with python 2.3
by Alex July 26, 2003

July 26, 2003

Hi I try to build boost.python with Visual C++ 7 using the build project given in ..\build\VisualStudio\ The build is ok when I use the python 2.2 includes and libs paths but I get something like 1600 errors when using python 2.3... have I something in particular to do to use python 2.3 ? regards Alex Legrand

2 1

Re: Problem build Boost.Python
by David Abrahams July 25, 2003

July 25, 2003

The following message is a courtesy copy of an article that has been posted to comp.lang.python as well. Bren <iambrenNOSPAM(a)sympatico.ca> writes: > Hi, > > Pardon my stupid, but I'm having a problem building Boost.Python as > per the instructions at > http://www.boost.org/libs/python/doc/building.html > > I got boost-build-2.0-m6.zip from sourceforge and build bjam.exe. It is possible to build Boost.Python under Boost.Build-2.0, but not using those instructions. Boost.Build v1 is still the official build system of Boost for the time being. [Boost.Build guys, do we need to rename the BBv2 prerelease so this confusion doesn't recur?] It looks like you're on windows, so you should either get the precompiled executable in bjam-3.1.4-2-ntx86.zip, or visit the tools/build/jam_src subdirectory of your Boost installation and invoke build.bat as per http://www.boost.org/tools/build/jam_src/index.html#installing > I've tried several ways: > > bjam -a > bjam -sTOOLS="msvc" > bjam -sTOOLS=msvc > > I tried with and without the using msvc ; in user-config.jam. > > In every case I get: > > C:/Project/boost/tools\boostbook.jam:86: in boostbook.init > *** argument error > * rule path.make ( native ) > * called with: ( ) > * missing argument native > C:/Project/boost/new\path.jam:42:see definition of rule 'path.make' > being called > > C:/Project/boost/new\toolset.jam:22: in using > C:\Project\boost\new\../new\user-config.jam:58: in modules.load > C:/Project/boost\build-system.jam:55: in load > C:\Project\boost\new\..\kernel\modules.jam:296: in import > C:\Project\boost\new\..\kernel\bootstrap.jam:122: in boost-build > C:\Project\boost\new\boost-build.jam:2: in module scope > > > Can anyone help? > HTH, -- Dave Abrahams Boost Consulting www.boost-consulting.com

1 0

pure virtual methods with pyste
by christoph.wiedemann＠daimlerchrysler.com July 25, 2003

July 25, 2003

Hello, i'm a boost.python / pyste newbie, and what i've seen so far looks very well to me. I am especially interested in using pyste for creating python bindings. However, i've ran into the following problem with pyste: All pure virtual methods of abstract classes are hidden in python, i.e. it's not accessible to use them in python. I've played around with 'normal' boost.python, and managed it to create wrappers for pure virtual methods. Is it planned to produce bindings for pure virtual functions or am i asking for the impossible ? Thanks for the great work ! Christoph

2 1

building Boost 1.30 with gcc 3.2.2 on Linux (Mandrake 9.1), w/Python 2.3
by Alex Martelli July 25, 2003

July 25, 2003

I'm trying -- and failing miserably. According to bjam -d+2 -a -q (run from boost_1_30_0) the C++ compilation command being used is: g++ -c -Wall -ftemplate-depth-100 -DBOOST_PYTHON_DYNAMIC_LIB -DBOOST_PYTHON_SOURCE -g -O0 -fno-inline -fPIC -I"libs/python/build" -I "/usr/local/include/python2.3" -I "/home/alex/pyr/boost_1_30_0" -o "libs/python/build/bin/libboost_python.so/gcc/debug/runtime-link-dynamic/shared-linkable-true/numeric.o" "libs/python/build/../src/numeric.cpp" This first warns about a redefinition of _POSIX_C_SOURCE (in /usr/local/include/python2.3/pyconfig.h:844:1 , already defined in /usr/include/features.h:131:1 -- hopefully a bening warning...?). Then the killers - a slew of error messages about /home/alex/pyr/boost_1_30_0/boost/python/converter/builtin_converters.hpp:106: of "parse error before & token" followed by "syntax error before & token", then again the same two messages for three more tokens (const > }), then again for lines 107, then for line 112 (and also 113, 114, 115, ...) errors that "to_python_value" is not a template, "detail" is not a class or namespace, etc etc. I've tried peering into the macros and templates involved but so far haven't come anywhere close to any understanding. I hope it's just some silly error on my part -- any suggestions...? Thanks, Alex

2 1

RE: [C++-sig] def_readonly and const array members
by Jeff Brewer July 25, 2003

July 25, 2003

I am using Boost 1.30. Sorry I forgot to mention that. def_readonly seems to still in 1.30 and its implementation is: template <class D, class B> self& def_readonly(char const* name, D B::*pm_) { D T::*pm = pm_; this->add_property(name, make_getter(pm)); return *this; } Is it deprecated though? However, I tried add_property with make_getter (which I usually use anyway) and I get the same error. Here is my new class_ code: class_<testStruct>("testStruct", init<>()) .add_property("theArray", make_getter(&testStruct::theArray, return_value_policy<return_by_value>())); Thank you for the help, Jeff Brewer jeff(a)brewer.com -----Original Message----- From: Ralf W. Grosse-Kunstleve [mailto:rwgk@yahoo.com] Posted At: Thursday, July 24, 2003 6:06 PM Posted To: c++-sig Conversation: [C++-sig] def_readonly and const array members Subject: Re: [C++-sig] def_readonly and const array members --- Jeff Brewer <jeff(a)brewer.com> wrote: > I am trying to wrap structs that have character arrays and I'm having > problems since I moved from MSVC 7 to MSVC 7.1 (VS .NET 2003). Here is > the struct I'm trying to expose the theArray member of using > class_::def_readonly: Are you also moving to a more recent boost? > .def_readonly("theArray", &testStruct::theArray); Try this instead: #include <boost/python/return_value_policy.hpp> #include <boost/python/return_by_value.hpp> .add_property("theArray", make_getter(&testStruct::theArray, return_value_policy<return_by_value>())); If you want to support both Boost 1.29 and never versions: #include <boost/version.hpp> #if BOOST_VERSION >= 103000 // use add_property #else // use def_readonly or def_readwrite #endif Ralf __________________________________ Do you Yahoo!? Yahoo! SiteBuilder - Free, easy-to-use web site design software http://sitebuilder.yahoo.com _______________________________________________ C++-sig mailing list C++-sig(a)python.org http://mail.python.org/mailman/listinfo/c++-sig

3 2

more cross-module problems
by greg Landrum July 25, 2003

July 25, 2003

Dave's patch cleared up the exception problems I reported yesterday, but I've got more cross-module problems to report. The attached zip file will create a directory called RDTest. If you extract it in $Boost/libs/python, you'll end up with something that can be built with bjam. I apologize in advance for the names of things here... The Jamfile runs two tests, both of which use multiple modules and explore passing objects between them. One of the tests works on Windows and fails on Linux, the other works on Linux and fails on Windows. The first test (named both_mods in the Jamfile) explores what happens if classes with boost::any data members are passed between modules. This test works fine on Windows but fails on Linux with the error: RuntimeError: boost::bad_any_cast: failed conversion using boost::any_cast The second test (named CM2_test) creates three modules (modulea, moduleb and modulec). modulea and modulec define simple classes (classA and classC) that use STL containers to map std::strings -> integers. classA does this with an std::map and classC does it with an std::vector of std::pairs. As the test demonstrates, both these modules work ok by themselves. moduleb has 2 functions: testCrossA() takes a classA instance and gets a property from it; testCrossC() takes a classC instance and gets a property. moduleb works without problems on Linux, but the test using classA instances (which carry around std::maps) fails on Windows with the dread: RuntimeError: unidentifiable C++ exception error. Platform notes; Windows = [Win2K, MSVC v6, python 2.2, boost 1.30.0] Linux = [RH8.0, g++ 3.2, python 2.2, boost 1.30.0] Thanks, -greg

3 4

FW: [C++-sig] Injected constructors
by Roman Yakovenko July 24, 2003

July 24, 2003

Just a question what is wrong with static member function or global function make_xxx? The idiom of constructing object and then making additional initialization is already used in Boost.Python - borrowed function for constructing handle. Also in Python code it will be written explicit what programmer is going to do. Did I missed something? Roman. -----Original Message----- From: David Abrahams [mailto:dave@boost-consulting.com] Sent: Wednesday, July 23, 2003 5:49 PM To: c++-sig(a)python.org Subject: [C++-sig] Injected constructors It's already possible to inject new methods into a Python class wrapper that are not in the underlying class: #include <vector> struct X { std::vector<int> v; }; int size(X& self) { return self.v.size(); } BOOST_PYTHON_MODULE(test) { class_<X>("X") .def("size", size) ; } >>> x = X() >>> x.size() 0 Currently, there's no easy way to do the same thing for constructors, without intruding on the underlying class or writing a wrapper class which repeats *all* of the base constructors and adds new ones. I am about to add just such a facility: std::auto_ptr<X> x_factory(PyObject* self, int n) { std::auto_ptr<X> x(new X); x->v.resize(n); return x; } BOOST_PYTHON_MODULE(test) { class_<X>("X") .def("size", size) .def("__init__", init_factory(x_factory)) ; } >>> x = X(30) >>> x.size() 30 init_factory creates a python callable object which calls its function argument and injects the resulting object into a holder in the self argument. I'm not in love with the "init_factory" name. Anyone have a better idea? -- Dave Abrahams Boost Consulting www.boost-consulting.com _______________________________________________ C++-sig mailing list C++-sig(a)python.org http://mail.python.org/mailman/listinfo/c++-sig

3 3