- Boost-users - lists.preview.boost.org

Boost/Python: method overloading w/out default args
by Bruce Lowery <bruce_lowery＠yahoo.com> 15 Jan '03

15 Jan '03

Using Boost/Python, how does one overload methods? For instance: struct Bar { void func( const std::string& s ); void func( int i ); }; BOOST_PYTHON_MODULE( Foo ) { class_<Bar>( "Bar" ) .def( "func", &Bar::func ) .def( ??? ) ; } The reference documentation discusses the BOOST_PYTHON_MEMBER_FUNCTION_OVERLOADS macro, but it only seems to apply to methods with default arguments. --bruce

2 2

Boost/Python: returning internal reference
by Bruce Lowery <bruce_lowery＠yahoo.com> 15 Jan '03

15 Jan '03

I have a simple C++ class that I've wrapped using Boost/Python, v1.29.0 (on Win 2K), and that has a simple accessor returning a const ref to an internal string. In the python interpreter, the accessor raises an exception (see below),whereas another access that returns a copy of the string works. Could someone explain my mistake? --bruce ------ Bar.cpp -------- #include <string> class Bar { public: Bar( const std::string& s ) : m_s( s ) {} std::string get_s_copy() const { return m_s; } const std::string& get_s_ref() const { return m_s; } std::string m_s; }; #include <boost/python.hpp> using namespace boost::python; BOOST_PYTHON_MODULE(Foo) { class_<Bar>( "Bar", init< const std::string& >( args("s") ) ) .def( "get_s_copy", &Bar::get_s_copy ) .def( "get_s_ref", &Bar::get_s_ref, return_internal_reference<>() ) .def_readonly( "m_s", &Bar::m_s ) ; } ----- Python session ---- Python 2.2.2 (#37, Oct 14 2002, 17:02:34) [MSC 32 bit (Intel)] on win32 Type "help", "copyright", "credits" or "license" for more information. >>> import Foo >>> bar = Foo.Bar("hello") >>> bar.m_s 'hello' >>> bar.get_s_copy() 'hello' >>> bar.get_s_ref() Traceback (most recent call last): File "<stdin>", line 1, in ? TypeError: bad argument type for built-in operation >>>

2 3

Boost/Python: Solaris build errors
by Bruce Lowery <bruce_lowery＠yahoo.com> 15 Jan '03

15 Jan '03

In building the boost/python library on solaris using v6.2.p2 of the sunpro compiler, two compile-time errors occur, one in 'cast.hpp' and the other in 'inheritance.hpp'. Details are below. --bruce ----------------- 1 -------------------- boost/boost/python/cast.hpp: line 65: Error: Default arguments cannot be added in later declarations of the template function in the same scope. 57 template <class Target, class Source> 58 inline Target* downcast(Source* p, yes_convertible) 59 { 60 return static_cast<Target*>(p); 61 } 62 63 template <class Target, class Source> 64 inline Target* downcast(Source* p, no_convertible, boost::type<Target>* = 0) *** 65 { *** 66 typedef typename base_type_traits<Source>::type base; 67 return (Target*)detail::downcast<base>(p, convertible<Source*>::check((base*)0)); 68 } ----------------- 2 -------------------- boost/boost/python/object/inheritance.hpp: line 134: Error: ")" expected instead of ">". 130 template <class Source, class Target> 131 inline void register_conversion( 132 // We need this parameter because CWPro7 can't determine 133 // which is the base reliably. *** 134 bool is_downcast = !cast_generator<Source,Target>::is_upcast *** 132 133 // These parameters shouldn't be used, they're an MSVC bug workaround 134 , Source* = 0, Target* = 0) 135 { 136 typedef typename cast_generator<Source,Target>::type generator; 137 138 add_cast(python::type_id<Source>() 139 , python::type_id<Target>() 140 , &generator::execute 141 , is_downcast); 142 }

2 2

RE: [Boost-Users] [BGL] missing include in boost/graph/kruskal_mi n_spanning_tree.hpp
by bjorn.karlsson＠readsoft.com 13 Jan '03

13 Jan '03

> From: Jeppe N. Madsen [mailto:yg-boost-users@m.gmane.org] > > Hi, > > kruskal_min_spanning_tree uses bgl_named_params defined in > boost/graph/named_function_params.hpp, which is not included.... Fixed in CVS. Thanks! Bjorn

1 0

BGL: filtered graphs, write_graphviz, and out of memory messages
by sa3h1n <sa3h1n＠yahoo.com> 13 Jan '03

13 Jan '03

Hi all: I am seeing a strange interaction between filtered graphs and write_graphviz. Briefly, if I try to write out a filtered graph via write_graphviz, compilation ends with an out-of-memory message. A necessary condition for this to happen is that the value_type of the property map I am filtering cannot be of a native type -- int is ok, std::string is not. Example code is appended below. I am using Codewarrior Dev Tools 8.0 on WinXP Pro. Given that the code does not even compile, I am not sure how to move beyond this point. Any ideas? TIA, - S ----------------------------------------------------------------- #include "boost/config.hpp" #include <fstream.h> #include <iostream> // for std::cout #include <utility> // for std::pair #include <algorithm> // for std::for_each #include "boost/utility.hpp" // for boost::tie #include "boost/graph/graph_traits.hpp" // for boost::graph_traits #include "boost/graph/adjacency_list.hpp" #include "boost/graph/graphviz.hpp" #include "boost/graph/filtered_graph.hpp" using namespace boost; namespace boost { enum edge_prop_t { edge_prop = 389457 }; BOOST_INSTALL_PROPERTY(edge, prop); } template <typename T> struct positive_edge_prop { positive_edge_prop() {}; positive_edge_prop(T t) : m_prop(t) {}; template <typename Edge> bool operator() (const Edge& e) const { return "foo" == get(m_prop, e); } T m_prop; }; int main(void) { // create an edge property // if std::string is replaced with int, and some attendant changes are made, the code compiles typedef property<edge_prop_t, std::string> EdgeProperty; // create a simple graph type typedef adjacency_list<vecS, vecS, bidirectionalS, no_property, EdgeProperty> Pgraph; enum { A, B, C, D, E, F, N }; const int num_vertices = N; typedef std::pair<int, int> Edge; Edge edge_array[] = { Edge(A, B), Edge(B, C), Edge(C, D), Edge(D, E), Edge(E, F), Edge(F, A), }; const int num_edges = sizeof(edge_array)/sizeof(edge_array[0]); Pgraph pg(edge_array, edge_array + num_edges, num_vertices); // Add edge properties typedef boost::property_map<Pgraph, edge_prop_t>::type ewt; ewt ew = get(edge_prop_t(), pg); boost::graph_traits<Pgraph>::edge_iterator ei, ei_end; boost::tie(ei, ei_end) = edges(pg); for (ei; ei != ei_end; ei++) put(ew, *ei, "bar"); std::map<std::string,std::string> graph_attr, vertex_attr, edge_attr; boost::write_graphviz(std::cout, pg); positive_edge_prop<ewt> filter(get(edge_prop_t(), pg)); filtered_graph<Pgraph, positive_edge_prop<ewt> > fg(pg, filter); boost::write_graphviz(std::cout, fg); // if this line is commented out, the code compiles just fine return 0; } -----------------------------------------------------------------

1 0

Next release date???
by Alejandro Escalante Medina 13 Jan '03

13 Jan '03

Does anyone have an idea for the date of the next release??? Thanks, Alex _________________________________________________________ Do You Yahoo!? La mejor conexión a internet y 25MB extra a tu correo por $100 al mes. http://net.yahoo.com.mx

2 1

Problem with g++ 2.95.3 and boost-example file_dependencies.cpp
by Michael Brinkmann 12 Jan '03

12 Jan '03

Hi, I am trying to get the "file_dependencies"-example compiled and running with the g++-version mentioned above. But, in the function "dijkstra_shortest_paths_no_init()" an exception is thrown and the program terminates prematurely. When hunting the bug with gdb, I get the message "Cannot access memory at address ..." for the parameter "weight", perhaps this is the reason for the exception. Is this a known problem? Will an upgrade to a higher g++-version help me, or has someone another hint? Thanks in advance Michael

1 0

threads - using boost.threads in win32 without use of dll?
by Phil Nash 12 Jan '03

12 Jan '03

Hi, I understand that on the win32 platforms boost.threads requires a dll to be build - threadmon. I seem to remember that this is there as a workaround to help support thread local storage (but I can't seem to find mention of it in the docs now, so I am not 100% sure of this). If this is the case, is it possible to use boost.threads *without* the dll if thread local storage is not being used? Regards, [)o IhIL..

1 0

boost::signal and thread-safety
by svanechel <svanechel＠yahoo.com> 11 Jan '03

11 Jan '03

Hi, The signal FAQ says that currently boost::signals aren't thread-safe. I'm guessing this applies to connecting/disconnecting slots from different threads and calling a signal from multiple threadsl, right? If synchronize access when connecting slots and make sure that only a single thread calls a certain signal, can I then use boost::signal to fire events from objects that do processing in a worker thread? As a secondary question: how far in the future will this thread- safety support be added? Thanks in advance, Sven

2 2

[BGL] missing include in boost/graph/kruskal_min_spanning_tree.hpp
by Jeppe N. Madsen 11 Jan '03

11 Jan '03

Hi, kruskal_min_spanning_tree uses bgl_named_params defined in boost/graph/named_function_params.hpp, which is not included.... rgds Jeppe

1 0