March 2010 - Boost-users - lists.preview.boost.org

[graph] Problem compiling simple graph code with recent g++
by Stephen Kell 18 Mar '10

18 Mar '10

I'm having some problems with the Boost Graph library in 1.42. Specifically, if I include the graph concepts header before declaring my overloads of vertices, out_edges et al, the compiler can't see my overloads. I've produced a small-ish test case (below) which doesn't compile with g++ 4.4.3. (It's not supposed to link...) In file included from test.cpp:3: /home/srk31/opt/include/boost/graph/graph_concepts.hpp: In destructor ?boost::concepts::VertexListGraph<G>::~VertexListGraph() [with G = Gr]?: /home/srk31/opt/include/boost/graph/graph_concepts.hpp:166: instantiated from ?static void boost::concept::requirement<Model>::failed() [with Model = boost::concepts::VertexListGraphConcept<Gr>]? /home/srk31/opt/include/boost/concept_check.hpp:43: instantiated from ?void boost::function_requires(Model*) [with Model = boost::concepts::VertexListGraphConcept<Gr>]? test.cpp:49: instantiated from here /home/srk31/opt/include/boost/graph/graph_concepts.hpp:187: error: no matching function for call to ?vertices(Gr&)? /home/srk31/opt/include/boost/graph/graph_concepts.hpp: In member function ?void boost::concepts::VertexListGraph<G>::const_constraints(const G&) [with G = Gr]?: /home/srk31/opt/include/boost/graph/graph_concepts.hpp:189: instantiated from ?boost::concepts::VertexListGraph<G>::~VertexListGraph() [with G = Gr]? /home/srk31/opt/include/boost/graph/graph_concepts.hpp:166: instantiated from ?static void boost::concept::requirement<Model>::failed() [with Model = boost::concepts::VertexListGraphConcept<Gr>]? /home/srk31/opt/include/boost/concept_check.hpp:43: instantiated from ?void boost::function_requires(Model*) [with Model = boost::concepts::VertexListGraphConcept<Gr>]? test.cpp:49: instantiated from here /home/srk31/opt/include/boost/graph/graph_concepts.hpp:201: error: no matching function for call to ?vertices(const Gr&)? What's odd is that if I move the #include of boost/graph/graph_concepts.hpp to *below* my overloads, it does work. However, that code reordering is not an option in the actual context where I'm having this problem. Can anyone see anything wrong with my code, or verify that they do or don't see the same problem? I'm thinking it's a compiler bug but am not sure... will file it in gcc bugzilla if I'm not proved wrong here. It does build successfully with g++ 4.1.2. If anyone knows a workaround, or has other suggestions, I'd be very grateful. Anyway, thanks for reading! Stephen #include <utility> #include <boost/graph/graph_traits.hpp> #include <boost/graph/graph_concepts.hpp> struct Gr { }; namespace boost { template <> struct graph_traits <Gr> { typedef void *vertex_descriptor; typedef std::pair<void*, void*> edge_descriptor; typedef void **vertex_iterator; typedef directed_tag directed_category; typedef allow_parallel_edge_tag edge_parallel_category; struct traversal_tag : public virtual vertex_list_graph_tag, public virtual incidence_graph_tag {}; typedef traversal_tag traversal_category; typedef size_t vertices_size_type; typedef size_t edges_size_type; typedef size_t degree_size_type; typedef std::pair <void*, void*> *out_edge_iterator; }; typedef std::pair<graph_traits<Gr>::out_edge_iterator, graph_traits<Gr>::out_edge_iterator> edge_iter_pair; typedef std::pair<graph_traits<Gr>::vertex_iterator, graph_traits<Gr>::vertex_iterator> vertex_iter_pair; edge_iter_pair out_edges (graph_traits<Gr>::vertex_descriptor u, const Gr& g); graph_traits<Gr>::degree_size_type out_degree( graph_traits<Gr>::vertex_descriptor u, const Gr & g); graph_traits<Gr>::vertex_descriptor source( graph_traits<Gr>::edge_descriptor e, const Gr & g); graph_traits<Gr>::vertex_descriptor target( graph_traits <Gr>::edge_descriptor e, const Gr & g); vertex_iter_pair vertices (const Gr & g); graph_traits<Gr>::vertices_size_type num_vertices(const Gr & g); } // end namespace boost int main () { Gr g; boost::out_edges (*boost::vertices (g).first, g); boost::function_requires < boost::IncidenceGraphConcept < Gr > >(); boost::function_requires < boost::VertexListGraphConcept < Gr > >(); }

2 2

[Interprocess] create a managed_mapped_file without allocator mutex?
by Andy Wiese 18 Mar '10

18 Mar '10

I'm using managed_mapped_file to hold an index for a small property database. In various operations of boost ::interprocess::rbtree_best_fit<boost::interprocess::mutex_family, boost::interprocess::offset_ptr<void>, 0ul> the process frequently blocks on boost::interprocess::interprocess_mutex::lock I synchronize all access to this file using interprocess mutex utilities, so write access is restricted to one single-threaded process at a time. I have looked for a deadlock between my outer mutex and the allocator's mutex, but I haven't been able to detect any process blocking on my mutex. It also might be that a process crashes while holding the allocator mutex, but in that case I don't know how to flush the allocators mutex (like I can with my own). I'm wondering if I need the allocator mutex at all, since I provide synchronization at a higher level. Is it possible to create a: typedef basic_managed_mapped_file< char,rbtree_best_fit< mutex_family >,iset_index > managed_mapped_file; That uses some kind of no-op mutex family? Can anyone give me a hint how to declare such a thing? Thanks much for any advice, Andy

2 2

Query Regarding mixing two different version of boost with same c++ application at same time
by akhilesh singh 18 Mar '10

18 Mar '10

Hi Boost users, Can we mix two different version of boost library(boost 1.34.1 and boost 1.39.0) with same appliaction at the same time? and if not, whether it will cause issue? Any help will be apprciated. Thanks Akhil

1 0

How are template arguments to rbtree_best_fit resolved
by Eric Whitcombe 18 Mar '10

18 Mar '10

I am stepping through interprocess shared memory code and I was a little confused by the offset_ptr code that I was hitting. I cannot figure out how that is specified to the template of rbtree_best_fit. I see in boost/interprocess/interprocess_fwd.hpp declaring managed_shared_memory by typedef'ing basic_managed_shared_memory with rbtree_best_fit<mutex_family> as the AllocationAlgorithm argument but I can't figure out how offset_ptr is is supplied as the VoidPointer argument to the rbtree_best_fit template. That template has 3 arguments but the forward declaration only provides one. Are the other two somehow inferred? If so, how?

2 1

Copy MPL sequence at runtime?
by Rhys Ulerich 18 Mar '10

18 Mar '10

Evening all, I'd like to copy the contents of a boost::mpl::vector_c (or similar) into an output iterator at runtime. I've not found a clean way to get an MPL sequence into a runtime-usable form. So far, I've gotten the included ugliness to work. 0) Is there some simple, idiomatic way to accomplish this goal that I've missed? 1) Any suggestions for how to clean this up? Specifically ways to reduce the assumptions about the underlying MPL sequence or fixes for braindead mistakes I've made? 2) Is there some way to have syntax like 'seqcopy<S>(iter)' where the iterator type is deduced but the sequence type is explicit? Thanks, Rhys #define BOOST_TEST_MODULE $Id: test_mpl.cc 996 2010-03-18 07:04:23Z rhys $ #include <boost/mpl/empty.hpp> #include <boost/mpl/front.hpp> #include <boost/mpl/pop_front.hpp> #include <boost/mpl/vector_c.hpp> #include <boost/utility.hpp> #include <boost/test/included/unit_test.hpp> using boost::disable_if; using boost::mpl::empty; using boost::mpl::front; using boost::mpl::pop_front; template<typename S, typename FI, class Enable = void> struct seqcopy { void operator()(FI fi) {} }; template<typename S, typename FI> struct seqcopy<S, FI, typename disable_if<typename empty<S>::type>::type> { void operator()(FI fi) { *fi++ = front<S>::type::value; seqcopy<typename pop_front<S>::type, FI>()(fi); } }; BOOST_AUTO_TEST_CASE( testing ) { typedef boost::mpl::vector_c<int,2,4,6> values; int a[3]; seqcopy<values, int*>()(a); BOOST_CHECK_EQUAL(a[0], 2); BOOST_CHECK_EQUAL(a[1], 4); BOOST_CHECK_EQUAL(a[2], 6); }

4 4

Problem Using managed_shared_memory
by med ennemri 18 Mar '10

18 Mar '10

Hi all I have a problem using managed_shared_memory , I can only read strings that has size less that 16 characters. This is my server code ////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// #include <boost/interprocess/managed_shared_memory.hpp> #include <utility> #include <iostream> int main () { using namespace boost::interprocess; typedef std::pair<std::string, int> FileState; try{ shared_memory_object::remove("MySharedMemory"); managed_shared_memory segment(create_only, "MySharedMemory", 655360); std::string str1 = "aaaaaaaaaaaaaaabbb";//client can only read <16 sized strings std::string str2 = "baaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"; std::string str3 = "caaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"; std::string *instance1 = segment.construct<std::string>("1")(str1); std::string *instance2 = segment.construct<std::string>("2")(str2); std::string *instance3 = segment.construct<std::string>("3")(str3); while(segment.get_num_named_objects() > 0) std::cout<<"Memory is ready"<<std::endl; } catch(...){ shared_memory_object::remove("MySharedMemory"); throw; } shared_memory_object::remove("MySharedMemory"); return 0; } ////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// Here is my client code ////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// #include <boost/interprocess/managed_shared_memory.hpp> #include <cstddef> #include <utility> #include <cassert> #include <iostream> #include <fstream> int main (int argc, char *argv[]) { using namespace boost::interprocess; try{ managed_shared_memory segment(open_only, "MySharedMemory"); std::pair<std::string*, std::size_t> res; int a = 1; std::ostringstream oss; oss<<a; res = segment.find<std::string> (const_cast<char*>(oss.str().c_str())); if(res.second != 0) { std::cout<<*(res.first)<<std::endl; segment.destroy<std::string>(const_cast<char*>(oss.str().c_str())); } } catch(...){ shared_memory_object::remove("MySharedMemory"); throw; } shared_memory_object::remove("MySharedMemory"); return 0; } ////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// Best Regards

2 2

Weird problem with boost::noncopyable and boost::mutex
by Daniel V 18 Mar '10

18 Mar '10

Hi, *I was doing a program in C++ where I simulate a Java's ArrayList...* * * template <class T> class Duo { public: Duo *next; T element; Duo(T el,Duo *sig) { element=el; next=sig; } ~Duo() {}; Duo* getNext() {return next;} T getElement() {return element;} void setNext(Duo *myNext) {next=myNext;} void setElement(T el){element=el;} }; template <class T> class ArrayList { public: Duo <T>*Ptr_First; Duo <T>*Ptr_Last; ArrayList(){Ptr_First = Ptr_Last = NULL;}; ~ArrayList(){} void insert(T el); std::pair<T, int> get(int index); }; template <class T> void ArrayList<T>::insert(T elem) { Duo<T> *ptr_walker = Ptr_First; Duo<T> *ptr_cache; if (Ptr_First == NULL && Ptr_Last == NULL) { Duo<T> *Ptr_temp = new Duo<T>(elem, NULL); Ptr_First = Ptr_temp; Ptr_Last = Ptr_temp; } else { while (ptr_walker->getNext() != NULL) ptr_walker = ptr_walker->getNext(); ptr_cache = ptr_walker; Duo<T> *Ptr_temp = new Duo<T>(elem, NULL); Ptr_Last = Ptr_temp; ptr_walker->setNext(Ptr_temp); } } template <class T> std::pair<T, int> ArrayList<T>::get(int index) { Duo<T> *ptr_walker; std::pair <T, int> receiver; receiver.first = ptr_walker->getElement(); receiver.second = 0; bool found = false; if ((Ptr_First == NULL) && (Ptr_Last == NULL)) return receiver; else { ptr_walker = Ptr_First; for (int i = 0; i<index; i++) { ptr_walker = ptr_walker->getNext(); } receiver.first = ptr_walker->getElement(); receiver.second = 1; return receiver; } } *The thing is that I instanciated an ArrayList object and called the "insert" method and I received an error from boost::mutex AND boost::noncopyable!! and I am not using them at all!!* *This is the code that generates the error:* ArrayList<ObjectA> myArrayListA; ObjectA myObjectA; myArrayListA.insert(myObjectA); *and the errors I got:* mkdir -p build/Debug/GNU-MacOSX rm -f build/Debug/GNU-MacOSX/main.o.d g++ -pthreads -lboost_thread -c -g -fPIC -MMD -MP -MF build/Debug/GNU-MacOSX/main.o.d -o build/Debug/GNU-MacOSX/main.o main.cpp /usr/local/include/boost/noncopyable.hpp: In copy constructor 'boost::mutex::mutex(const boost::mutex&)': In file included from main.cpp:23: /usr/local/include/boost/noncopyable.hpp:27: error: 'boost::noncopyable_::noncopyable::noncopyable(const boost::noncopyable_::noncopyable&)' is private /usr/local/include/boost/thread/pthread/mutex.hpp:31: error: within this context ObjectA.h: In copy constructor 'ObjectA::ObjectA(const ObjectA&)': ObjectA.h:18: note: synthesized method 'boost::mutex::mutex(const boost::mutex&)' first required here *Any Ideas why am I receiving this errors?* THANKS!! Dann

3 4

[multi_index] Is replace() ok within container iteration loop?
by PB 18 Mar '10

18 Mar '10

Hello, This question is about boost:multi_index_container. Code is iterating over a multi_index_container and replacing some elements. The data being replaced has nothing to do with any keys/indices. Is this allowed by multi_index_container? Thanks, Pete

2 2

[asio] async_read_until'smatch_condition requirements/guarantees
by Maxim Yanchenko 18 Mar '10

18 Mar '10

Hi Chris/All, I didn't find anything about this in the docs, so asking here. Why I'm asking: for some protocols matching condition can be pretty complex and will include some parsing of a partially received message. It would be great to reuse information that was already parsed, to avoid parsing same things many times. Also, when we finally received the whole message and match_condition returns true, the handler is invoked which will do mostly the same parsing that was already made by the match_condition - this is also a thing that would be great to avoid. 1) Is the same match_condition object used every time, or multiple copies can be created? 2) Does the state of the match_condition object persist between different condition checks? 3) How match_condition is used if we have several threads? Should I protect access to the state with this regard? 4) In which thread match_condition is executed? is it internal asio thread or one of the threads that called io_service.run()? 5) What if I call the handler code right in the match_condition if the condition is met? In this case I can reuse data I already extracted while checking the condition, and will also avoid waiting for my handler in the asio queue. 6) Same as 5), with one addition - what if I have only one thread that calls io_service.run()? Of course I can figure out all this stuff from the source code, but it's obviously subject to change, so I would better rely on something more stable, like docs (would be great to eventually have answers to the above questions in the docs). Thanks, Maxim

1 0

[boost-users] [assign] std::vector initialisation by boost::assign::list_of
by Andrey Torba 18 Mar '10

18 Mar '10

The code is compilable on msvc but is not on gcc-4.3: #include <boost/assign/list_of.hpp> #include <vector> int main() { //OK std::vector<int> v = boost::assign::list_of(1)(2); // error std::vector<int> v(boost::assign::list_of(1)(2)); return 0; } I can initialise std::vector in a first way. But the problem occurs when i try to initialise std::vector in a class initialization list: class sss { std::vector<int> v; sss() : v(boost::assign::list_of(1)(2)) {} }; -- Regards, Andrey

2 7