October 2009 - Boost-users - lists.boost.org

[Asio] when stopping io_service, I get a system_error : mutex: invalid argument
by philipp_muenzel＠web.de 03 Oct '09

03 Oct '09

Hello, I'm using asio for a udp receiver. I used the example code from http://www.boost.org/doc/libs/1_40_0/doc/html/boost_asio/example/multicast/… and created a class that in its contructor opens a socket, binds to it, calls async_receive_from, and in the receive handler again calls async_receive from. Now i have manager class that takes care of the io service, the thread, and an instance of the receiver class. Everything works fine until the moment I destroy the manager class. Here is what the destructor looks like: Manager::~Manager { // first, make sure that no further requests are handled that would prevent io_service from quitting m_receiver->getSock()->shutdown(boost::asio::ip::udp::socket::shutdown_both); m_receiver->getSock()->close(); // destroy the receiver by resetting its auto_ptr m_receiver.reset(); // reset the work auto_ptr, destroying the work for io_service, so that it can quit m_io_work.reset(0); // join the thread in which io_service.run() was called. m_receiver_thread.join(); // now stop the io // this line is commented, it has no effect whether i remove the slashes // m_io_service.stop(); } When I destroy the manager, i get the following error: "terminate called after throwing an instance of 'boost::exception_detail::clone_impl<boost::exception::detail::error_info_injector<boost::system::system_error> > ' what(): mutex: Invalid argument" It seems similar to the problem that was discussed here: http://lists.boost.org/Archives/boost/2008/03/134169.php But I'm pretty sure that given I destroy the instance that holds the socket first, the io_service has no work left. I tried catching the exception within the destructor and within the run() of the thread, but no luck. I cannot find the place where the exception is thrown, nor do I have an idea what else to do to ensure the io_service has no work left. Any help on fixing this is greatly appreciated. Regards, Philipp ________________________________________________________________ Neu: WEB.DE Doppel-FLAT mit Internet-Flatrate + Telefon-Flatrate für nur 19,99 Euro/mtl.!* http://produkte.web.de/go/02/

1 0

[Process (Vault)] child::wait error on OS X 10.6
by Rush Manbert 03 Oct '09

03 Oct '09

I am using the version of the Process library that was announced by Boris Schaeling in this document: http://www.highscore.de/cpp/process/ I used the link in the introduction to get the zip file, so I think I have the correct version. I am running the following test function: namespace bp = boost::process; namespace bios = boost::iostreams; int BoostProcessTest (void) { std::cout << "\nTesting launch of child process using Boost.Process\n"; std::string exec = bp::find_executable_in_path("svn"); std::vector<std::string> args; args.push_back("svn"); args.push_back("info"); // context bp::context ctx; // child launch bp::child c = bp::launch(exec, args, ctx); // sleep(1); //wait const bp::status s = c.wait(); // parse status if (s.exited() && s.exit_status() == EXIT_SUCCESS) std::cout << "Directory is a working copy." << std::endl; else std::cout << "Directory is not a working copy." << std::endl; std::cout << std::endl; return 0; } I call this function in a loop for as long as it returns 0, which should be forever. The test is run in a directory that is NOT a svn working copy, so all that the child process does is to write a line to stderr and exit. What I'm seeing, however is this message: Error: Caught an exception! boost::process::child::wait: waitpid(2) failed: Interrupted system call The only way I have found to make that go away is by adding the sleep (1) call (commented out above). Then it runs forever. I'm hoping Boris will read this and have some idea of what's happening. My google searches haven't turned up much of value. - Rush

2 1

[Graph] Type "VertexIndexMap"
by Alexandre Bos 03 Oct '09

03 Oct '09

Hello, I am trying to create an instance of edmonds_augmenting_path_finder : --------- test.cpp --------- #include <boost/graph/adjacency_list.hpp> #include <boost/graph/max_cardinality_matching.hpp> using namespace std; using namespace boost; typedef adjacency_list<vecS, vecS, undirectedS> custom_graph; [...] typedef std::vector<graph_traits<custom_graph>::vertex_descriptor> matemap_t; matemap_t mate(42); custom_graph g(42); edmonds_augmenting_path_finder<custom_graph, matemap_t, property_map<custom_graph, vertex_index_t>::type > augmentator(g, mate, get(vertex_index, g)); [...] --------------------------- The compiler (g++) returns the following error : /.../max_cardinality_matching.hpp: In constructor 'boost::edmonds_augmenting_path_finder<Graph, MateMap, VertexIndexMap>::edmonds_augmenting_path_finder(const Graph&, MateMap, VertexIndexMap) [with Graph = boost::adjacency_list<boost::vecS, boost::vecS, boost::undirectedS, boost::no_property, boost::no_property, boost::no_property, boost::listS>, MateMap = std::vector<long unsigned int, std::allocator<long unsigned int> >, VertexIndexMap = boost::vec_adj_list_vertex_id_map<boost::no_property, long unsigned int>]': /.../convert.cpp:102: instantiated from here /.../max_cardinality_matching.hpp:190: error: no matching function for call to 'get(std::vector<long unsigned int, std::allocator<long unsigned int> >&, const long unsigned int&)' How can I fix this ? Thank you, Alexandre

2 2

[mpl] Discrepancy between documentation and code?
by James Knight 03 Oct '09

03 Oct '09

In looking at the documentation for the intrinsic metafunction mpl::at, it seems to suggest that a three argument version that returns a default type when a key is not found exists for associative sequences. I don't see this implemented anywhere in the code, and the simple example below doesn't compile. Am I missing something or is this an error in the documentation? Thanks, Nate #include <boost/mpl/map.hpp> #include <boost/mpl/at.hpp> #include <boost/mpl/pair.hpp> #include <boost/mpl/assert.hpp> #include <boost/mpl/int.hpp> #include <boost/type_traits/is_same.hpp> using namespace boost::mpl; int main() { typedef map< pair<int_<2>,int_<3> > > map_t; typedef at<map_t, int_<1>, int_<0> >::type missing; BOOST_MPL_ASSERT(( boost::is_same< missing, int_<0> > )); } With gcc 4.0.1 and Boost 1.37 on OS X I get test.cpp: In function 'int main()': test.cpp:12: error: wrong number of template arguments (3, should be 2) ../BoostInstall/boost_1_37_0/boost/mpl/at_fwd.hpp:20: error: provided for 'template<class Sequence, class N> struct boost::mpl::at' ...

1 0

problems with program_opsitons::parse_environment can't figure out how to do what I want.
by Caleb Cushing 02 Oct '09

02 Oct '09

variables_map vm_env; store(parse_environment(desc, "HOME"), vm_env); notify(vm_env); if(vm_env.count("HOME")) { path rcfile = ".koramarc"; path home; home = vm_env["HOME"].as<path>(); path tmp = home / rcfile; cout << "check" << endl; // just a retarted test to see if we're actually reaching the end of this... } -- Caleb Cushing http://xenoterracide.blogspot.com

2 3

Can't compile some boost code on Solaris
by Roger House 02 Oct '09

02 Oct '09

I'm attempting to build a quite large app on Solaris (which I know next to nothing about), and I'm getting compilation errors in a couple of boost files, seemingly related to stlport. The error message "exception is not a member of stlp_std" is issued for line 29 of archive_exception.hpp: class archive_exception : public virtual std::exception The same message (plus others which I think are spurious) is issued for line 55 of throw_exception.hpp: inline void throw_exception_assert_compatibility( std::exception const & ) { } All this code compiles okay with gcc on Linux and with cl on Windows, so I think it is specifically a Sun issue. A Google search led to some discussions on the boost site, but I could not determine from them what the fix is. The Sun compiler I'm using is version 5.9.0, and the boost version is 1.39.0. Any help on this will be greatly appreciated. Roger House Software Developer

1 0

How to use the shared region in append mode
by manish4gupta 02 Oct '09

02 Oct '09

First of all i would like to thank for previous answers. Now i am able to use the shared region to create a map, insert value into shared map and use the shared map. Now moving one step ahead. I am planning to use the shared region in append mode. For instance 1. create shared region if it is not used before // how to check this ? managed_shared_memory segment(create_only,MySharedMemory, 999999999); 2. create map in the region // done 3. open the map and use the map for retrieving and inserting values //done How i can that the given region already allocated to map so that i can insert the value in it otherwise i can create the that region and create tha map and insert the value in the map. Thanks in advance. -- View this message in context: http://www.nabble.com/How-to-use-the-shared-region-in-append-mode-tp2567740… Sent from the Boost - Users mailing list archive at Nabble.com.

2 5

[function/bind] function object type introspection
by Gennadiy Rozental 02 Oct '09

02 Oct '09

Hi, I've got couple questions. Might be someone can point me in right direction. Let's say I've got this class ConcreteFunctor which should handle all kinds of functions. I want to store the function as boost::function object, thus the class has template parameter - FunctionType: template<typename FuncType> ConcreteFunctor : public AbstractFunctor { ConcreteFunctor( FuncType f ) ... {} virtual invoke( std::vector<Object*> const& args ) { // here I can use FuncType::arity and FuncType's argument types // to properly convert arguments to desired types and invoke m_f } FuncType m_f; } Now I need to implement object maker template<typename F> boost::shared_ptr<AnstractFunctor> make_functor( F f ) { typedef boost::function<???> TypeName; return boost::shared_ptr<AbstrctFunctor>( new ConcreteFunctor<FuncType>( f ) ); } I was able to work fine with free functions by just removing pointer from F. Now to the questions: 1. How do I deduce boost::function type from pointer to member functions: Essentially if F is R (A::*)(T1,T2) I need to get boost::function<T (A*,T1,T2)> 2. This is more difficult: if f is bound object, how to I intrspect it's type: Essentially in use case like this: int foo( std::string, int ) {...} make_functor( boost::bind( &foo, "abc" ) ) I want to get FuncType as boost::function<int (int)> Regards, Gennadiy

3 5

[asio] async connect vs sync connect
by Mathieu Peyréga 02 Oct '09

02 Oct '09

Hello, I have a TCP client program using boost 1.40 asio framework and a question. I have an external LAN hosted TCP server ready to accept connexions (functionnal and tested with other means) On my client side, when using the async connect function, it sometimes take up to 2-3 seconds before the connexion handler is called. If I use the synchronous connect function, it is almost immediate. Is there a way to speed up the connexion time, while using the async method ? Regards, Mathieu

1 0

[asio] deadline_timer makes wrong delay when using date command
by Bjarne Laursen 02 Oct '09

02 Oct '09

I have used a deadline_timer to generate a polling every 1 second. Unfortunely when I use the linux date command the timing is curupted. (If I set the clock back 1 min. I will be missing 60 pollings). What can I do about that? -Bjarne

4 9