- Boost-users - lists.boost.org

Interprocess : offset_ptr used as array
by Oodini 16 May '13

16 May '13

Hello, I have an offset_ptr named tmpData defined like this : offset_ptr<const byte>* someVariable; indextype index; ... offset_ptr<const byte> tmpData = ((offset_ptr<const byte>*)someVariable)[index]; data points actually to the beginning of an array. The original code, without shared memory was : byte* data= ((byte**)m_someVariable)[index]; The problem is that I have to translate that : indextype* data = (indextype*)&(tmpData[1]); The adress of the second element is taken though the & adress. But when tmpData is an offset_ptr, the [] operator returns a reference, and I can apply the & operator. Is there any solution that copying the pointer to temporary variable, and increment this one ? Thanks.

1 0

Re: [Boost-users] boost::variant & c++11
by Antony Polukhin 16 May '13

16 May '13

2013/5/15 Bryan Catanzaro <bcatanzaro(a)acm.org>: > Hi Antony - > Perhaps it is compiler dependent. Which version of MSVC did you reproduce > the error with? Does that version have c++11 support enabled (in particular, > rvalue references) > As I said, with Boost 1.48.0, this code works with g++ 4.6 and 4.7. With > 1.53.0, it works with g++ 4.6 and 4.7 only if c++11 support is turned off. > If c++11 support is turned on, the code does not work. > I've attached a shell script that compiles the attached code with the cross > product of {(g++4.6, g++4.7), (boost1.48, boost1.53), (noc++11, yesc++11)}. > As you can see from test.txt, the error happens only with boost 1.53 and > c++11 support. > > > - bryan Exactly the same things happen on GCC-4.5 and Clang-3.0. But what is more interesting here: convertible c; const convertible& cc = c; // Always compiles, using any flags, compiler and Boost version my_variant y = cc; my_variant y1 = static_cast<const convertible&>(c); my_variant y2 = static_cast<const convertible>(c); // Always fail to compile, using any flags and Boost version // my_variant z = c; // my_variant y3 = static_cast<convertible&>(c); Tested on clang, gcc4.5, gcc 4.6, gcc4.7 on Boost 1.48 and Boost 1.53 with and without C++11 flags. If we modify `convertible`: struct convertible { operator my_variant() { return my_variant(); } operator my_variant() const { return my_variant(); } }; Then ALL the tests pass well, even the commented out ones. Even `my_variant y4 = static_cast<convertible&&>(c);` works well. I'll test it tomorrow on Visual Studio 11 once more and investigate issue futher. -- Best regards, Antony Polukhin

2 2

PROBLEM WITH BOOST MPI
by siva teja 15 May '13

15 May '13

Hi friends, I have to run distributed_delta_stepping with graphs in CSR fromat. So I tried with "distributed_shortest_path_test.cpp" located at "<top of boost source tree>/libs/graph_parallel/test/ ". It reads a random graph in adjacency list format. But I have to run DIMACS_USA graph in CSR format. So, I edited the code but it resulted in following errors. I guess I am facing problems while representing a graph in CSR format.. CODE : void test_distributed_shortest_paths(int n, double p, int c, int seed) { char *graf = "./dimacs_USA_sort_wdup" ; typedef compressed_sparse_row_graph<directedS, VertexProperties, WeightedEdge, no_property, distributedS<mpi_process_group> > Graph; typedef graph_traits<Graph>::vertex_descriptor vertex_descriptor; typedef graph_traits<Graph>::edge_descriptor edge_descriptor; typedef graph_traits<Graph>::vertex_iterator vertex_iterator; typedef graph_traits<Graph>::vertices_size_type vertices_size_type; typedef property_map<Graph, vertex_index_t>::type vertex_index_map; typedef std::pair<long int, long int> Edge; struct timeval starttime,endtime,timediff; gettimeofday(&starttime,0x0); char ch; long int num_nodes; long int num_arcs; Edge *edge_array; long int *weights; long int arc_index=0, wt_index=0; long int vert1, vert2, dist; std::fstream input; input.open(graf); if( !input.eof()) { input >> ch; if(ch == 'vertex') { input >> num_nodes; } } if(!input.eof()) { input >> ch; if(ch =='edge') { input >> num_arcs; } else std::cout << "Wrong graph format\n"; } edge_array=(Edge *)malloc(sizeof(Edge) * num_arcs); weights=(long int *)malloc(sizeof(long int) * num_arcs); while(!input.eof()) { input >> ch; if(ch == 'arc') { input >> vert1 >> vert2 >> dist; Edge temp; temp.first=vert1; temp.second=vert2; edge_array[arc_index++]=temp; weights[wt_index++]=dist; } } gettimeofday(&endtime, 0x0); Graph g(edge_array, edge_array + num_arcs, weights, num_nodes); vertex_descriptor s = vertex(1, g); graph::distributed::delta_stepping_shortest_paths(g, s, dummy_property_map(), get(&VertexProperties::distance, g), get(&WeightedEdge::weight, g)); } But when I run the command " mpirun -np 1 ./a.out" I face the following errors. **** Testing aborted. **** 1 error detected ------------------------------ -------------------------------------------- mpirun has exited due to process rank 0 with PID 19096 on node rise-02 exiting without calling "finalize". This may have caused other processes in the application to be terminated by signals sent by mpirun (as reported here). -------------------------------------------------------------------------- Thank you in advance.

1 0

Statechart timer extension and new version of visualizer
by Petr Silhavik 15 May '13

15 May '13

Hi, I have implemented timers for library Boost Statechart. It was tested only on Linux because it uses POSIX threads. It is extension that replaces fifo_scheduler and fifo_worker with only one class Scheduler. That means that it is suitable only for asynchronous state machines. The source codes with example and short description can be downloaded here: https://www.dropbox.com/s/ubtep8f7hs6g5xl/boostTimers.tar.gz I would like to also inform you about new version of Boost Statechart viewer. Now it is plugin for compiler Clang and it was tested vitw version 3.0 and higher - (mostly tested using version 3.1). You can find more information and news about it on the web page of this project where you can also download the source codes and see some examples of the output pictures. http://rtime.felk.cvut.cz/statechart-viewer/ Best regards Petr

2 2

boost::variant & c++11
by Bryan Catanzaro 15 May '13

15 May '13

Hi - I seem to have run into a regression with boost::variant from 1.48 to 1.53. With C++11 enabled, using g++, the following code seems to confuse variant's constructor selection machinery: if you look at the line after #ifndef WAR, the implicit conversion isn't happening, and so boost::variant attempts to convert construct from extract<my_variant> to A or B. It should instead copy construct from the output of extract<my_variant>'s implicit conversion to my_variant. If I force an explicit conversion, this problem doesn't occur. If I don't enable c++11, this problem doesn't occur. With Boost 1.48, this problem doesn't occur. Any advice from the list as to why this behavior changed? Thanks, bryan #include <boost/variant.hpp> //A simplified wrapper template<typename T> struct wrap { T t; wrap(T& _t) : t(_t) {} operator T() const { return t; } T force() const { return t; } }; struct A{}; struct B{}; typedef boost::variant<A, B> my_variant; int main() { A a; my_variant x(a); #ifndef WAR my_variant y = wrap<my_variant>(x); #else my_variant y = wrap<my_variant>(x).force(); #endif } And the error: /home/bcatanzaro/boost_1_53_0/boost/variant/variant.hpp:1574:9: error: no matching function for call to ‘boost::variant<A, B>::initializer::initialize(void*, std::remove_reference<wrap<boost::variant<A, B> >&>::type)’

3 3

Constraining unordered template constructors
by Michael Marcin 15 May '13

15 May '13

I have a class that I want to have accept 0-5 arguments in any order for it's constructor. Each argument is a const references to one of 5 possible class types. All arguments must be distinct in type, passing the same type twice in an error. Passing an argument that is not one of the 5 supported types is also an error. How do you go about implementing something like this? I figure first you have to create a constructor for each arity but then how do you constrain the constructors as described above? I imagine it would involve enable_if. class widget { widget(); template< class A > widget( const A& a ); template< class A, class B > widget( const A& a, const B& b ); template< class A, class B, class C > widget( const A& a, const B& b, const C& c ); template< class A, class B, class C, class D > widget( const A& a, const B& b, const C& c, const D& d ); template< class A, class B, class C, class D, class E > widget( const A& a, const B& b, const C& c, const D& d, const E& e ); };

4 3

directory monitoring with dynamically added dirs
by Richard Langly 14 May '13

14 May '13

I'm trying to dynamically add directories from a config file before I enter my programs main loop using the ASIO dir monitor example. I've not been able to understand the internals yet. Below is my current iteration of what I'm trying. I have been able to get the simple examples running just like in the async.cpp file. But I can't get adding dirs dynamically working and was hoping someone could offer suggestions. Any help much appreciated. void my_file_handler(const boost::system::error_code &ec, const boost::asio::dir_monitor_event &ev) { // event stuff ... return; } boost::asio::dir_monitor *dm; boost::asio::io_service io_service; vector<boost::asio::dir_monitor *> vecdm; for (vector<Directories>::iterator it = dirs->begin(); it != dirs->end(); ++it) { dm = new boost::asio::dir_monitor(io_service); dm->add_directory((*it).abs_dir_path); vecdm.push_back(dm); } boost::thread t = boost::thread(boost::bind(&boost::asio::io_service::run, boost::ref(io_service))); while (true) { // MAIN PROGRAM LOOP for (vector<boost::asio::dir_monitor *>::iterator it = vecdm.begin(); it != vecdm.end(); ++it) { (*it)->async_monitor(my_file_handler); } io_service.reset(); // DO PROGRAM STUFF ... } ~ ~

1 0

[msm] Injecting events to FSMs from different translation units
by samriti katoch 14 May '13

14 May '13

Hi, I am using boost msm to design state machines for a relatively large project. The documentation online is pretty good and I was able to figure out most features by looking at the sample code (thanks!). My particular application has multiple FSM instances active at any given time, and what I would like to do is to be able to generate events for a particular FSM instance from different translation units. But I'd like to keep the the compile time at a minimum. I found an answer from Christophe for a similar question sometime back here: http://stackoverflow.com/questions/10899641/boost-msm-class Since, I have multiple FSM instances it would not help to have a singleton class with the FSM shared_ptr. I am looking for ideas/suggestions/experiences for implementing this in an elegant OO ways. Thanks, Samriti

3 2

CANNOT COMPILE BOOST 153 WITH GCC-4.8
by Didier J-P Guion Firmin 14 May '13

14 May '13

Hi guys, I've tried to compile boost 153 with gcc-4.8 and two weird things are happening. 1 - a bunch of warning during compile time 2 - any program using boost thread errors out : Threading support unavaliable: it has been explicitly disabled with BOOST_DISABLE_THREADS" Can you please help me figure out what I'm doing wrong Thank you --

4 6

boost::addressof and std::nullptr_t
by Mathieu Champlon 12 May '13

12 May '13

Hello, With MSVC 2010 the following code fails to compile : std::nullptr_t np; boost::addressof( np ); The error being : 1>....\boost/utility/addressof.hpp(59): error C2666: 'boost::detail::addressof_impl<T>::f' : 2 overloads have similar conversions 1> with 1> [ 1> T=std::nullptr_t 1> ] 1> ....\boost/utility/addressof.hpp(43): could be 'std::nullptr_t *boost::detail::addressof_impl<T>::f(T *,int)' 1> with 1> [ 1> T=std::nullptr_t 1> ] 1> ....\boost/utility/addressof.hpp(37): or 'std::nullptr_t *boost::detail::addressof_impl<T>::f(T &,long)' 1> with 1> [ 1> T=std::nullptr_t 1> ] 1> while trying to match the argument list '(boost::detail::addr_impl_ref<T>, int)' 1> with 1> [ 1> T=std::nullptr_t 1> ] 1> ....blabla.cpp(356) : see reference to function template instantiation 'T *boost::addressof<std::nullptr_t>(T &)' being compiled 1> with 1> [ 1> T=std::nullptr_t 1> ] Can this be considered a bug in boost::addressof ? As a side note it compiles with std::addressof which implementation for MSVC 2010 is simply : template<class _Ty> inline _Ty * addressof(_Ty& _Val) { // return address of _Val return ((_Ty *) &(char&)_Val); } Thank you, MAT.

1 0