- Boost-users - lists.boost.org

[boost.python] exec_file crashes when using python 3.2
by yjaeger＠free.fr 04 Jul '12

04 Jul '12

Hi, I'm very new to boost, and I first want to thank you all for the hard and impressive work that has been done. I'm currently trying to use Boost.python and I'm facing an annoying issue. The following source code compiles fine (with msvc10 and mingw gcc 4.7): #include <boost/python.hpp> #include <iostream> using namespace boost::python; void greet() { // Retrieve the main module. object main = import("__main__"); // Retrieve the main module's namespace object global(main.attr("__dict__")); // Define greet function in Python. object result = exec("def greet():\n" " return 'Hello from Python!' \n", global, global); // Create a reference to it. object greet = global["greet"]; // Call it. std::string message = extract<std::string>(greet()); std::cout << message << std::endl; } void greet_file() { // Retrieve the main module. object main = import("__main__"); // Retrieve the main module's namespace object global(main.attr("__dict__")); // Define greet function in Python. object result = exec_file("hello.py", global, global); // Create a reference to it. object greet = global["greet"]; // Call it. std::string message = extract<std::string>(greet()); std::cout << message << std::endl; } void main(int argc, char** argv) { Py_Initialize(); greet(); greet_file(); } It even runs perfectly well when I'm using boost.python linked against python 2.7.3 library. I have following output: Hello from Python! Hello from Python! The issue appears when I link boost.python against python 3.2 (tried with 3.2.2 and 3.2.3) the first 'Hello from Python!' shows up, but then it crashes and all I can see is a windows message box saying 'prog.exe has stopped working'. It's not that I absolutely want to use python 3, but as I'm working on a new project, I'm concerned with durability, so I would prefer 3.2. Any idea would be welcome. Thanks for reading. Yannick.

1 0

[bgl] depth_first_search does not compile. color_map howto?
by Joachim Faulhaber 04 Jul '12

04 Jul '12

Hi, When using visitors with the depth_first_search algorithm I can successfully compile and run this code: //========================================================= template<class Graph> class Visitor: public default_dfs_visitor { public: typedef typename graph_traits<Graph>::vertex_descriptor Vertex; void discover_vertex(Vertex v, const Graph& g)const {cout << v << " "; return;} }; void GLV_visit() { typedef adjacency_list<listS, vecS, directedS> GraphLV; GraphLV g; add_edge(0, 1, g); add_edge(0, 2, g); add_edge(1, 2, g); add_edge(1, 3, g); Visitor<GraphLV> vis; boost::depth_first_search(g, visitor(vis)); cout << endl; } BOOST_AUTO_TEST_CASE(graph_test) { GLV_visit(); } //========================================================= With a slightly different graph (listS instead of vecS for vertices) the same call of depth_first_search fails to compile with msvc-10. //========================================================= struct Int{ Int(): _value(0){} Int(int val): _value(val){} int _value; }; void GLL_visit() { typedef adjacency_list<listS, listS, directedS, Int> GraphLL; typedef graph_traits<GraphLL>::vertex_descriptor VertexLL; GraphLL g; Int val(42); VertexLL v0 = boost::add_vertex(g); VertexLL v1 = boost::add_vertex(g); g[v0] = Int(0); g[v1] = Int(1); add_edge(v0, v1, g); Visitor<GraphLL> vis; boost::depth_first_search(g, visitor(vis)); //compile errors } //========================================================= >From the tiny error message of 108 long verbose lines of internal graph template instantiations (see attachment) it seems that the compiler is unable to create a default color_map (but I might be wrong with this interpretation of the template jungle). 1. Do I need to provide a color_map parameter here? 2. If so, what is the simplest way to do that and provide an object of the appropriate type? I'd greatly appreciate a working code example. Examples and best practices for defining and handling external propery maps in general would be helpful. TIA, Joachim -- Interval Container Library [Boost.Icl] http://www.joachim-faulhaber.de

2 5

Re: [Boost-users] [boost.test] Missing entry point when statically linking
by Tan, Tom (Shanghai) 04 Jul '12

04 Jul '12

> Date: Tue, 3 Jul 2012 19:25:50 +0000 (UTC) > From: Gennadiy Rozental <rogeeff(a)gmail.com> > >Tan, Tom (Shanghai <TTan <at> husky.ca> writes: > > #include <boost/test/unit_test.hpp> > using namespace boost::unit_test::framework; using namespace > boost::unit_test; ? > test_suite* init_unit_test_suite(int argc, char *argv[]) { ??? > BOOST_TEST_MESSAGE("hello test."); ? > ??? return 0; > } > With the same build options: > cl.exe /EHsc /MT main.cpp > ? > VC2010 compiles fine. VC2012 outputs: > LINK : fatal error LNK1561: entry point must be defined. >Don't you need to link with static library? Can you try using single header variants? >Gennadiy I am trying to use the static library, because I was under the impression that the feature in this variant would be more complete than others. With VC++ I do not have to explicitly specify the static file name in the command line if it's what you mean. And if this is the problem, the error would be like LINK : fatal error LNK1104: cannot open file 'libboost_unit_test_framework-vc110-mt-s-1_50.lib' I had a re-read of the single header variant documentation. Seems I can just switch to it. But it still seems strange that the static variant works with VC10 but not VC11.

2 1

Re: [Boost-users] [boost.test] Missing entry point when statically linking
by Tan, Tom (Shanghai) 04 Jul '12

04 Jul '12

>Date: Tue, 3 Jul 2012 19:52:31 +0100 >rom: Marco Craveiro <marco.craveiro(a)gmail.com> > >Hi Tom, > >shouldn't it be > >#include <boost/test/included/unit_test.hpp> > >on the include? Thanks, Marco. I was intending to use the library variant because I recall reading somewhere the header only variant missing some test tools. Judging from this link: (http://www.boost.org/doc/libs/1_50_0/libs/test/doc/html/utf/user-guide/ usage-variants/single-header-variant.html) I just read, I seem to be wrong. The code in question was in place since around 2005. I never thought of changing it until it's rejected by vc11. Thanks for the suggestion.

1 0

Hello, I am using boost graph, I have a questions about disable edge in one graph
by mingliang zou 03 Jul '12

03 Jul '12

I created a graph. And use add_edge to add some edges, I wanna disable some edges, and add some edges. and use boost graph algorithm for enable edges How can I do this? not create new graph.

2 5

Compilation Error related to 'range'
by VimalMathew＠Eaton.com 03 Jul '12

03 Jul '12

Hi, I have this line of code in my header file: project(C, range(0, A_rows), range(0, A_cols)) = A; I've specified: #include <boost/numeric/ublas/storage.hpp> using namespace boost::numeric::ublas; at the beginning of the header file When I compile my .cpp file, I get an error message saying: "c:\users\c9995799\desktop\level 1 problem for mpi\dmatrix.hpp(34): error C2784: 'const boost::numeric::ublas::matrix_range<M> boost::numeric::ublas::project(const boost::numeric::ublas::matrix_range<M> &,const matrix_range<M>::range_type &,const matrix_range<M>::range_type &)' : could not deduce template argument for 'const boost::numeric::ublas::matrix_range<M> &' from 'boost::numeric::ublas::matrix<T>'" This error goes away if I substitute boost::numeric::ublas::range instead of just range here: project(C, range(0, A_rows), range(0, A_cols)) = A; How do I get this to work the way it currently is? Thanks, Vimal

2 1

[boost.test] Missing entry point when statically linking to VC2012 but no problem with VC2010
by Tan, Tom (Shanghai) 03 Jul '12

03 Jul '12

I have been using a source template for unit test with boost.test, which used to be working well until I started to use VC2012 RC. Here's a minimal repo: #include <boost/test/unit_test.hpp> using namespace boost::unit_test::framework; using namespace boost::unit_test; test_suite* init_unit_test_suite(int argc, char *argv[]) { BOOST_TEST_MESSAGE("hello test."); return 0; } With the same build options: cl.exe /EHsc /MT main.cpp VC2010 compiles fine. VC2012 outputs: LINK : fatal error LNK1561: entry point must be defined. Added BOOST_TEST_MAIN caused errors for both compilers. error C2084: function 'boost::unit_test::test_suite *init_unit_test_suite(int,char *[])' already has a body I tested on both 1.49 and 1.50. And the results are the same. I suspect there is a bug of version check against MSVC but could not locate it.

3 2

boost-spirit: need help with simple parser
by Allan Nielsen 03 Jul '12

03 Jul '12

Hi I would like to use boost::spirit for parsing ipv4 addresses. Here is what I have tried to do: struct Ipv4 { Ipv4() { raw.as_int = 0; } Ipv4( const Ipv4& i ) { raw.as_int = i.raw.as_int; } Ipv4& operator=(const Ipv4& i ) { raw.as_int = i.raw.as_int; return *this; } Ipv4(const boost::tuple<uint8_t, uint8_t, uint8_t, uint8_t>& v) { raw.as_char[3] = v.get<0>(); raw.as_char[2] = v.get<1>(); raw.as_char[1] = v.get<2>(); raw.as_char[0] = v.get<3>(); } union { uint32_t as_int; uint16_t as_short[2]; uint8_t as_char[4]; } raw; }; struct DecOctet : qi::grammar<const char *, uint8_t()> { DecOctet() : DecOctet::base_type(start) { start %= qi::uint_parser<uint8_t, 10, 1, 3>(); } qi::rule<const char *, uint8_t()> start; } dec_octet; struct Ipv4Address : qi::grammar<const char *, Ipv4()> { Ipv4Address() : Ipv4Address::base_type(start) { start %= dec_octet >> qi::lit('.') >> dec_octet >> qi::lit('.') >> dec_octet >> qi::lit('.') >> dec_octet ; } qi::rule<const char *, Ipv4()> start; } ipv4_address; bool parse(const const_string& s, Ipv4& i) { Ipv4 _i; const char * iter = s.begin(); bool r = qi::parse(iter, s.end(), ipv4_address, _i); if( !r || iter != s.end() ) return false; i = _i; return true; } But unfortunately if I parse an address such as "1.2.3.4", I end up with the address "1.0.0.0" in my Ipv4 struct. The IPv4 struct is legacy code, and the union member can therefor not be changed. I can add other constructors, but I prefer non-intrusive changes. So here are my questions: - What have I done wrong? - How should this has been done if I did not have a constructor accepting the tuple? - How would you have done? Best regards Allan W. Nielsen

3 7

FW: Does the boost library free the memory occupied by the vector automatically when I call erase method provided by list class?
by Gordon HUYUE 03 Jul '12

03 Jul '12

From: huyuevip(a)hotmail.com To: boost-users(a)lists.boost.org Subject: Does the boost library free the memory occupied by the vector automatically when I call erase method provided by list class? Date: Tue, 3 Jul 2012 10:03:34 +0000 Hi all, Currently I am using "boost/interprocess" to manage shared memory between parent process and child process, the OS is AIX 6.1. The boost library is really cool and it works well, but I still have something to worry about, because the program is server side application and should be available 24/7. I apply memory at the beginning of the program, the size of memory is big enough; I use a list object provided by boost, the element type of the list object is a request object, the request object has a member with property of vector, Here is my concern: Does the boost library free the memory occupied by the vector automatically when I call erase method provided by list class? if it does not , the memory will be run out soon, because the program will add and delete object from list frequently. I'd like to make sure whether the boost library can deal with it. If it cann't, Is there any other solutions? some tips when reading demo file Line 55: definition of RequestObject Line 67: this is a member variable inside RequestObject interfaceobject_vector myvector; Line 79: definition of List, the element of this List is RequestObject Line 139: we add a request object into list. myRequestObjectList->push_front(*myRequestObject); Line 268: we delete a request objest from list, in this case , Does boost library free the memory occupied by myvector(which is a member variable inside RequestObject) automatically? myRequestObjectList->erase(myIterator); The file attached is a demo cpp, I hope it will be helpful for everybody to understand my question easily. If we add InterfaceObject_sh object into myvector when initializing, and add request object(which include myvector) into list, finally we call erase methed provided by list, Does boost library free the memory occupied by myvector automatically? Best Regards, Gordon

1 0

[Phoenix] Simple if_ issue
by Aaron_Wright＠selinc.com 02 Jul '12

02 Jul '12

I'm trying to do a simple if/else statement using boost::phoenix and I seem to mis-understanding something. namespace bp = boost::phoenix; namespace bpa = bp::arg_names; boost::function< std::string () > test1 = bp::val("true"); boost::function< std::string (bool) > test2 = bp::if_(bpa::arg1) [ bp::val("true") ] .else_ [ bp::val("false") ]; The first one works, but the second one doesn't. I get an error about not being able to convert void to std::string: cannot convert parameter 1 from 'void' to 'const std::basic_string<_Elem,_Traits,_Ax> & What do I need to do in the second test to get it to compile? Can I tell if_ that its return type is not void? --- Aaron Wright

3 3