April 2009 - Boost-users - lists.preview.boost.org

[BGL] unordered_set<std::vector<edge_descriptor> >
by Johan Oudinet 02 May '09

02 May '09

Hi, I want to keep a set of paths, which are implemented as a vector of edge_descriptor. I don't care about the order. I just have to add paths and be able to know the number of different paths present in the set. So, I plan to use unordered_set but I'd like to know you opinion about a hash function, since AFAIK, unordered_set is like an hash_set? My main issue is to use as few memory as possible, because there will have a lot of paths in the set. Any idea? Thanks for your help. -- Johan

3 5

performance of dijkstra_shortest_paths
by Bhaskara Marthi 01 May '09

01 May '09

I recently ran into a situation where dijkstra_shortest_paths in the boost graph library was running very slowly. When I replaced the call to dijkstra_shortest_paths with the code below, things were a couple of orders magnitude faster. Any idea why? - Bhaskara Version 1: with dijkstra_shortest_paths ReachableCostVector GridGraph::singleSourceCosts (const Cell2D& source, const vector<Cell2D>& dests) { GridDistances distances; GridGraphVertex start = cellVertex(source); resetIndices(); typedef map<GridGraphVertex, GridGraphVertex> GridPreds; GridPreds predecessors; boost::dijkstra_shortest_paths (graph_, start, weight_map(get(&EdgeCost::cost, graph_)). vertex_index_map(get(&CellInfo::index, graph_)). distance_map(associative_property_map<GridDistances>(distances)). predecessor_map(associative_property_map<GridPreds>(predecessors)). // to avoid warnings visitor(GraphSearchVisitor())); ReachableCostVector v(dests.size()); transform (dests.begin(), dests.end(), v.begin(), bind(extractCost, this, distances, _1)); return v; } Version 2: without dijkstra_shortest_paths // Approximation to Dijkstra ReachableCostVector GridGraph::singleSourceCosts (const Cell2D& source, const vector<Cell2D>& dests) { GridDistances distances; GridGraphVertex start = cellVertex(source); typedef pair<GridGraphVertex, double> QueueItem; queue<QueueItem> q; set<GridGraphVertex> visited; q.push(QueueItem(start,0.0)); visited.insert(start); while (!q.empty()) { GridGraphVertex v; double d; tie(v,d) = q.front(); q.pop(); GridGraphAdjacencyIterator iter, end; for (tie(iter,end)=adjacent_vertices(v, graph_); iter!=end; iter++) { if (visited.find(*iter)==visited.end()) { double cost = d+graph_[edge(v,*iter,graph_).first].cost; q.push(QueueItem(*iter, cost)); visited.insert(*iter); distances[*iter] = cost; } } } ReachableCostVector costs; for (vector<Cell2D>::const_iterator iter=dests.begin(); iter!=dests.end(); ++iter) { GridDistances::iterator pos = distances.find(cellVertex(*iter)); if (pos==distances.end()) costs.push_back(ReachableCost(false,-42.0)); else costs.push_back(ReachableCost(true, pos->second)); } return costs; }

4 7

[boost-users][StateChart] polymorphic_downcast failure
by Igor R 01 May '09

01 May '09

Hi, The above failure occurs in reaction_dispatcher::derived_with_action::react(). IIUC, state_machine tries to invoke in_state_reaction that's defined for an event, which is unrelated to the one being processed. What logical error in my state-machine might cause such a problem? Unfortunately, I don't have any self-contained code at the moment, so please treat my question as theoretical :). Thanks.

2 6

[statechart] Asynch statechart in dll
by Bill Clark 01 May '09

01 May '09

I have modified the Pingpong sample (an asynchronous statechart) so that the statechart itself is in a dll, with the calling code (in PingPong.cpp) in an exe. This is using VS2005 sp1. I have used a macro in the usual way to mark the Player class as exported. The app builds and runs, but it seems that events don't get acted on. We hit Player's initiate_impl(), and we get to state_machine::process_event() with a BallReturned event passed in, but we never hit the handler for that event (Waiting::react(const BallReturned&)). Is this something that should work? Is it necessary to do something more than just export the necessary classes from the dll? Thanks, Bill **

2 4

Overloading errors
by A B 01 May '09

01 May '09

Does anyone know why the following sample fails (it was adapted from the boost overloading example) and how to do this correctly. Thanks in advance. #include <boost/python/module.hpp> #include <boost/python/def.hpp> #include <boost/python/args.hpp> #include <boost/python/tuple.hpp> #include <boost/python/class.hpp> #include <boost/python/overloads.hpp> #include <boost/python/return_internal_reference.hpp> using namespace boost::python; struct X { int f(int x, double y = 4.25, char const* z = "wow") { return x; } int f(int x, bool b = true) { return x; } }; BOOST_PYTHON_MEMBER_FUNCTION_OVERLOADS(f_member_overloads, X::f, 1, 3) BOOST_PYTHON_MODULE(args_ext) { class_<X>("X", "This is X's docstring") .def("f1", &X::f, f_member_overloads(args("x", "y", "z"), "f's docstring" )[return_internal_reference<>()] ) ; } g++ -fpic -shared -o test.so test.cpp -lpython2.4 -lboost_python -I/usr/include/python2.4 test.cpp: In function ‘void init_module_args_ext()’: test.cpp:31: error: no matching function for call to ‘boost::python::class_<X, boost::python::detail::not_specified, boost::python::detail::not_specified, boost::python::detail::not_specified>::def(const char [3], <unresolved overloaded function type>, boost::python::detail::overloads_proxy<boost::python::return_internal_reference<1ul, boost::python::default_call_policies>, f_member_overloads>)’

2 2

Boost.MultiIndex question on merge
by arm2arm 01 May '09

01 May '09

Hello, Are there way to merge two multiindex objects ? suppouse we have : struct strData{ int id; float a, b; strData(int id_, float a_, float b_):a(a_),b(b_),id(id_){} }; typedef multi_index_container< strData, indexed_by< ordered_unique< tag<id>, BOOST_MULTI_INDEX_MEMBER(strData,int,id)>, ordered_non_unique< tag<snap>,BOOST_MULTI_INDEX_MEMBER(strData,float,a)> > > data_setA, data_setB; Are there way to do: data_setA.merge(data_setB) ? thank you beforehand. Arman. -- View this message in context: http://www.nabble.com/Boost.MultiIndex-question-on-merge-tp23277496p2327749… Sent from the Boost - Users mailing list archive at Nabble.com.

5 11

boost coroutine status
by Robert Ramey 30 Apr '09

30 Apr '09

This package has been in the vault or some time. At least one well known person on this list has indicated that the quality is good and that he is using it in production code. Turns out that I now have a need of this facility but I've got a couple of problems. a) When I try and build it, I get ConvertFiberToThread which looks to me that there is some bjam magic missing. b) I would like to look the documentation. But that's hard to do as it's in quick book files and I don't have a quick book reader. I don't know that I have the ability to generate html files from quickbook. Maybe I do, but I can't really figure this out. c) I don't think that this package has ever been reviews nor do I see it scheduled for review. What is the status of this library. Anyone having information on this topic is encouraged to bring me up to date. Robert Ramey

5 13

[lexical_cast] float-to-string rounding error (not found with cout)
by Grant Birchmeier 30 Apr '09

30 Apr '09

I'm seeing unexpected behavior when using lexical_cast<string>. This code: const float f = 00123.12300; cout << f << endl; cout << boost::lexical_cast<std::string>(f) << endl; gives the result: 123.123 123.123001 Two questions: (1) Is there a way to eliminate this trailing '1' with lexical_cast (presumably some wierd rounding error)? (2) When casting float-to-string, given this behavior, why should I use lexical_cast instead of, say, stringstream? Thanks -Grant

4 5

Default argument values with boost::function
by Artyom 30 Apr '09

30 Apr '09

Hi, Is there any way to set default argument values with Boost::Function? On VS2008 SP1 I tried: boost::function< void( unsigned i = 1 ) > increment; But this didn't work. Code such as increment( 1 ); compiles OK but if you try doing increment(); the following compiler error is triggered: error C2064: term does not evaluate to a function taking 0 arguments class does not define an 'operator()' or a user defined conversion operator to a pointer-to-function or reference-to-function that takes appropriate number of arguments Thanks, Artyom

5 4

[boost-users][python] Pointer argument failed conversion?
by Germán Diago 30 Apr '09

30 Apr '09

Hello. After much googling I didn't find anything useful. My problem is: I have a virtual function like this class MyClass { void monitor(Object * o); } I have wrapped another class, which inherits from Object: class Character : public Object {...}; which is also exposed to python. I register my virtual function as the documentation says I have to do: class_<MyClass, boost::noncopyable>("MyClass"). def("monitor", &MyClass::monitor); When I try to call it from python, I get the following error: a = MyClass() b = Character(....) a.monitor(b) MyClass.monitor(MyClass, Character) did not match C++ signature: monitorizar(MyClass {lvalue}, Object *) Shouldn't my Character be automatically converted to Object * and then to make a successful call to the function? Should I do anything special to use pointers as parameters of my functions? I don't know why it tries to pass a "Character". What I want is to get it converted to Object *. Thanks in advance.

2 1