May 2006 - Boost-users - lists.preview.boost.org

[boost-users] [parameter] test\preprocessor.cpp
by Cromwell Enage 01 Jun '06

01 Jun '06

With the latest CVS, test\preprocessor.cpp no longer compiles. The first two errors are: macro BOOST_PARAMETER_FOR_EACH_pred_aux3 requires 3 arguments, but only 2 given macro BOOST_PP_SPLIT_0 requires 2 arguments, but only 1 given All of my programs that use BOOST_PARAMETER_FUNCTION et al. are failing due to these errors. Cromwell D. Enage __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

2 5

boost::bind with const parameter
by Alain Cormier 01 Jun '06

01 Jun '06

Hi group! I try to call in a for_each a member function that take a parameter by value AND const. I know we do not need to do things like that, but I am using code from an other team and I don't want to change it if I don't need. It results with this error : D:\ENG-AC\working\tests\test_ParamConst_vs_boostBind\test\test.cpp(35) : error C2665: 'bind' : none of the 2 overloads can convert parameter 1 from type 'int (__thiscall test_class::*)(const int) const' Here a little program that reproduce de behavior : #include <algorithm> #include <vector> #include "boost/bind.hpp" class test_class { public: test_class(int value) : m_value(value){} ~test_class(){} int fct_param_const(const int value) const {return m_value + value;} int fct_param_ref_const(const int& value) const {return m_value + value;} int fct_param(int value) const {return m_value + value;} private: int m_value; }; int main(int argc, char* argv[]) { std::vector<int> all_values(5); std::generate(all_values.begin(), all_values.end(), &rand); test_class doTest(28); std::for_each(all_values.begin(), all_values.end(), boost::bind(&test_class::fct_param, doTest, _1)); std::for_each(all_values.begin(), all_values.end(), boost::bind(&test_class::fct_param_ref_const, doTest, _1)); // DOESN'T WORK : because of the parameter by value and const??? //std::for_each(all_values.begin(), all_values.end(), boost::bind(&test_class::fct_param_const, doTest, _1)); return 0; } Is this a correct behavior from boost::bind? Does the only solution is to remove the const from our function? We have so much difficulty to stay const-correct, I do not want to send a signal to our team that const isn't good (but maybe too much const isn't good :D). Anyway, what do you suggest? By the way. I am using VC6 (SP5) under Win2000 and boost 1.33.1. Regards, -- Alain Cormier alain.work(a)gmail.com

2 2

[Test] Is It Possible to Put Test Cases in DLL?
by Wu Yinghui, Freddie 31 May '06

31 May '06

Dear all, I remember seeing similar topic some time earlier, but I failed to find the information I need. So if this question has been answer before, please kindly point me to the correct resource. TIA! Basically, I goal is to build a bunch of unit test cases into a few DLLs, and be able to run them by loading the DLLs into my test executable (either automatic or using LoadLibrary is fine with me). I'm using the auto registration facility now to create and register the test suites and cases. And I'm running on MSVC 7.1 on Windows XP SP2 now. Any pointer will be greatly appreciated. Cheers, Freddie -- Wu Yinghui, Freddie Research & Development Software Engineer Volume Interactions Pte Ltd 1 Kim Seng Promenade, #12-01 Great World City East Tower Singapore 237994 Tel: +65 62226962 (Ext 216) Fax: +65 62226215 Email: yhwu(a)volumeinteractions.com URL: http://www.volumeinteractions.com Important: This message is intended for the recipient(s) addressed above. It contains privileged and confidential information. If you are not the intended recipient, please notify the sender immediately by replying to this message and then delete it from your system. You must not read, copy, use, or disseminate this communication in any form. Thank you.

3 2

Re: [Boost-users] [BGL] Updating to CVS version, typedef ... Graph not working?!
by dave＠boost-consulting.com 31 May '06

31 May '06

I think there's probably a better way: put all the concepts in a nested namespace of boost and then add a using-directive to expose them in boost... Although I'd really like to see the whole test program first, because it's not obvious where/how the collision occurs. -----Original Message----- From: dgregor(a)osl.iu.edu Subj: Re: [Boost-users] [BGL] Updating to CVS version,typedef ... Graph not working?! Date: Wed May 31, 2006 10:31 am Size: 927 bytes To: boost-users(a)lists.boost.org cc: dave(a)boost-consulting.com > I switched from 1.33.1 to CVS, and after that, my test programm did not > work any longer ... > > The relevant part is: > > typedef adjacency_list<vecS, vecS, undirectedS, > property<vertex_name_t, std::string, > property<decomposition_index_t, int, > property<coordinates_sq_t, simple_point<double>, > property<coordinates_circle_t, Polar<double>, > VParams > > > > > MyGraph; > > > > (Before it was Graph instead of MyGraph). [snip] > Can someone explain to me what's happening here? This looks like fallout from recent changes to the concept checking library. There is now a template called "Graph" in namespace boost, and we find that as well as your "Graph" typedef. This might force us to go back to the "Concept" suffix on concept checking templates, if this problem affects many compilers :( Doug -- Dave Abrahams Boost Consulting http://www.boost-consulting.com

3 2

[Graph] BGL python bindings documentation
by John Reid 31 May '06

31 May '06

Hi, Can anyone point me in the direction of some documentation of the python bindings for the BGL? I know this is no longer part of boost. Should I be asking some other list? The documentation at http://www.osl.iu.edu/~dgregor/bgl-python/#example and http://www.osl.iu.edu/~dgregor/bgl-python/reference/boost.graph.html is helpful but far from complete. What happened to the more user friendly documentation that came with boost_1_33_1? How stable is the interface? Is the library actively maintained? After all that I'd like to say I have found the bindings very useful just a little hard to work with at first. Thanks, John.

2 1

[BGL] Updating to CVS version, typedef ... Graph not working?!
by Jens Müller 31 May '06

31 May '06

I switched from 1.33.1 to CVS, and after that, my test programm did not work any longer ... The relevant part is: typedef adjacency_list<vecS, vecS, undirectedS, property<vertex_name_t, std::string, property<decomposition_index_t, int, property<coordinates_sq_t, simple_point<double>, property<coordinates_circle_t, Polar<double>, VParams > > > > > MyGraph; (Before it was Graph instead of MyGraph). The compiler error is not really enlightening to me: ens@fairtrade ~/c++ $ LC_ALL=en g++ -O3 my_fr.cpp my_fr.cpp:50: error: `Graph' was not declared in this scope my_fr.cpp:50: error: template argument 1 is invalid my_fr.cpp:50: error: `type' does not name a type my_fr.cpp:51: error: `Graph' was not declared in this scope my_fr.cpp:51: error: template argument 1 is invalid my_fr.cpp:51: error: `type' does not name a type my_fr.cpp:54: error: `Graph' was not declared in this scope my_fr.cpp:54: error: template argument 1 is invalid my_fr.cpp:54: error: `vertex_descriptor' does not name a type my_fr.cpp:76: error: `Graph' was not declared in this scope my_fr.cpp:76: error: template argument 2 is invalid my_fr.cpp:76: error: ISO C++ forbids declaration of `SWGen' with no type my_fr.cpp: In function `int main()': my_fr.cpp:83: error: use of `Graph' is ambiguous my_fr.cpp:49: error: first declared as `typedef class boost::adjacency_list<boost::vecS, boost::vecS, boost::undirectedS, boost::property<boost::vertex_name_t, std::string, boost::property<decomposition_index_t, int, boost::property<coordinates_sq_t, boost::simple_point<double>, boost::property<coordinates_circle_t, Polar<double>, VParams> > > >, boost::no_property, boost::no_property, boost::listS> Graph' here /usr/local/include/boost/graph/graph_concepts.hpp:31: error: also declared as `template<class G> struct boost::Graph' here my_fr.cpp:83: error: `Graph' was not declared in this scope my_fr.cpp:83: error: expected `;' before "g" my_fr.cpp:88: error: `CoordSqMap' was not declared in this scope my_fr.cpp:88: error: expected `;' before "position" my_fr.cpp:89: error: `g' was not declared in this scope my_fr.cpp:89: error: `position' was not declared in this scope Old program: #include <boost/graph/fruchterman_reingold.hpp> #include <boost/graph/random_layout.hpp> #include <boost/graph/adjacency_list.hpp> #include <boost/graph/simple_point.hpp> #include <boost/lexical_cast.hpp> #include <string> #include <iostream> #include <map> #include <vector> #include <boost/random/linear_congruential.hpp> #include <boost/progress.hpp> #include <boost/shared_ptr.hpp> #include <boost/graph/adjacency_list.hpp> #include <boost/graph/small_world_generator.hpp> #include <boost/random/linear_congruential.hpp> using namespace boost; struct decomposition_index_t{ typedef vertex_property_tag kind; }; struct coordinates_sq_t{ typedef vertex_property_tag kind; }; struct coordinates_circle_t{ typedef vertex_property_tag kind; }; template<typename T> struct Polar { T rad; T angle; }; struct VParams { int a; double b; long c; // std::complex<double> d; }; typedef adjacency_list<vecS, vecS, undirectedS, property<vertex_name_t, std::string, property<decomposition_index_t, int, property<coordinates_sq_t, simple_point<double>, property<coordinates_circle_t, Polar<double>, VParams > > > > > Graph; typedef property_map<Graph,coordinates_sq_t>::type CoordSqMap; typedef property_map<Graph,coordinates_circle_t>::type CoordCMap; typedef graph_traits<Graph>::vertex_descriptor Vertex; class progress_cooling : public linear_cooling<double> { typedef linear_cooling<double> inherited; public: explicit progress_cooling(std::size_t iterations) : inherited(iterations) { display.reset(new progress_display(iterations + 1, std::cerr)); } double operator()() { ++(*display); return inherited::operator()(); } private: shared_ptr<boost::progress_display> display; }; typedef boost::small_world_iterator<boost::minstd_rand, Graph> SWGen; int main() { boost::minstd_rand gen; // Create graph with 10000 nodes int nodecount = 4000; Graph g(SWGen(gen, nodecount, 6, 0.03), SWGen(), nodecount); double width = 2; double height = 2; CoordSqMap position = get(coordinates_sq_t(), g); random_graph_layout(g, position, -width/2, width/2, -height/2, height/2, gen); int iterations = 20; fruchterman_reingold_force_directed_layout (g, position, width, height, cooling(progress_cooling(iterations)).force_pairs(all_force_pairs() ) ); return 0; } Can someone explain to me what's happening here?

2 1

[Parameter] Positional arg after named arg
by Richard Crossley 31 May '06

31 May '06

Hi, With the parameter library is there any way to stop misuse as in; func(/*a = */1,b = 2); Fine, a is positional, b is named. func(a = 1,2); Happily compiles but probably isn't what was intended, the default value is used for the second parameter. Should positional arguments be allowed to occur after named arguments? Perusing the Parameter library docs I can't see any mention as to what should happen here. Regards, Richard. (using VC8.0 & cvs head)

2 2

[program_options]Lack of information in exceptions thrown
by Peter Gerell 31 May '06

31 May '06

Hi, I would like to know the name of the option that causes the command line parsing to fail. I don't like the idea of having to parse the what() string to extract the option name. I think the exception classes should be extended with functions to get this information. This is useful if you want to control the appearance of the error messages. E.g. use another language than English. In short I would like to do something like this: try { ... } catch ( const po::multiple_occurrences& e) { std::cerr << "multiple occurrences of option " << e.option_name(); } catch ( const po::unknown_option& e ) { std::cerr << "unknown option " << e.option_name(); } catch ( const po::invalid_syntax& e ) { std::cerr << e.what()); } catch ( const po::error& e ) { std::cerr << e.what()); } /Peter

1 0

[serialization] Overriding serialization of class pointer types
by Bryan Donlan 31 May '06

31 May '06

Hi all, I'm currently working on a game engine, and hoping to use boost serialization for the world saving code. It's nearly perfect except for one snag - serializing pointers to sprite file classes. Ideally I'd like to simply serialize the name of the sprite file, and on deserialization find the file in the sprite cache 'gallery', but this requires using split serialization in all the classes that reference sprites. Is there a way to simply override the serialization of /pointers/ to classes? I've tried the following: typedef creaturesImage *p_img; namespace boost { namespace serialization { template <class Archive> inline void save(Archive &ar, const p_img &obj, const int version) { bool isNull = (obj == NULL); ar & isNull; if (!isNull) { std::string name = obj->serializedName(); ar & name; } } template <class Archive> inline void load(Archive &ar, p_img &obj, const int version) { bool isNull; ar & isNull; if (isNull) obj = NULL; else { std::string name; ar & name; obj = world.gallery.getImage(name); } } } } BOOST_SERIALIZATION_SPLIT_FREE(creaturesImage *); BOOST_CLASS_IMPLEMENTATION(creaturesImage *, boost::serialization::primitive_type); However, this seems to be ignored; the compiler errors complain of such things as /usr/include/boost/serialization/access.hpp:109: error: 'class creaturesImage' has no member named 'serialize' (full log at http://pastebin.com/747527) The code in question can be found at http://ccdevnet.org/viewcvs.cgi/trunk/?rev=941 (ser/ subdirectory and serialization.h particularly), or by subversion at svn://ccdevnet.org/openc2e/trunk (rev 941 contains the code cited here; build with make openc2e_s; dependency list at http://openc2e.ccdevnet.org/download) Thanks, Bryan Donlan

2 3

program_options positional options and help message
by Pierre THIERRY 31 May '06

31 May '06

Hi, I'm just discovering po, and I wonder how to achieve the following: I want that the help message that I produce by streaming my options description to standard output also mentions the positional options, in the way that the traditional 'usage:' message does: usage: $PROG [options] FILE1 FILE2 ... Is it possible to do somethink like that? Am I clear? Curiously, Nowhere man -- nowhere.man(a)levallois.eu.org OpenPGP 0xD9D50D8A

2 1