- Boost-users - lists.boost.org

How is the proactor pattern of boost.asio implemented in linux?
by Journeyer J. Joh 10 Apr '13

10 Apr '13

Hello list, In ACE and on Linux the Proactor pattern is not recommended. Instead of this, ACE_Dev_Poll_Reactor seems to be recommended. It is one of Reactor implementation. ACE_Dev_Poll_Reactor uses epoll in a level-trigger method. But, I am not sure of this information.... I just found some blog essays on internet. Now I wonder about boost.asio. boost.asio must have a proactor implementation on Linux, I guess. How does the proactor of boost.asio work? Does it use epoll? Is it edge-triggered? What about real-time signal? Doesn't boost.asio use real-time signal on Linux? What is the recommended implementation of boost.asio for asynchronous and non-blocking overlapped IO on Linux? Thank you in advance! ---------------------------------------- Journeyer J. Joh o o s a p r o g r a m m e r a t g m a i l d o t c o m ----------------------------------------

1 0

ASIO with std::thread
by asif saeed 10 Apr '13

10 Apr '13

Hi, I am a newbie to both boost.asio and std::thread and std::future and so on and am just learning and trying to develop a messaging library. Are there any implications of using std::thread and std::future, etc. with boost.asio (instead of boost.thread, etc.)? I'd be thankful if you could provide me the answer in detail and give me helpful pointers to online links and provide examples. Specifically, I'm very interested in using std::future and promise with std::thread but I want future with ASIO based sockets. In addition, it'd be just great if you could also provide me an example of using boost's chrono based stead clock for setting up interval timers to, for example, get an event after every 10 seconds. Best regards, -Asif

1 0

Efficient "[][]" operator overloading (encapsulating a MultiArray)
by Jiří Vyskočil 09 Apr '13

09 Apr '13

Hello, I need to encapsulate a 3D MultiArray inside another class "GridArray". The reason is that GridArray will have another implementation not using MultiArray and it seems much easier to provide an abstration over MultiArray, than to reimplement its interface for the other implementation. I need to keep the [][][] operator access to the structure. I created a class like this, but it is slow: template<class T> class GridArray<T,3> { public: GridArray(int x, int y, int z): _array(boost::extents[x][y][z]) { } boost::multi_array<T, 3> _array; struct SliceRef { struct ColRef { SliceRef &_parent; int _x, _y; ColRef(SliceRef &p, const int x, const int y): _parent(p), _x(x), _y(y) {} T& operator[](const int z) { return _parent._parent._array[_x][_y][z]; } }; GridArray<T,3> &_parent; int _x; SliceRef(GridArray<T,3> &p, const int x): _parent(p), _x(x) {} ColRef operator[](const int y) { return ColRef(*this,_x,y); } }; SliceRef operator[](const int x){ return SliceRef(*this,x); } }; I am unable to write a minimal test case - if I compare using multi_array directly vs. using this GridArray class in a simple test program (compiled with -O2), the performance is the same. But the real code shows about 20% slowdown - it's a numerical simulaiton, I create several arrays at the beginning and then do computations with them. The rest of the code hasn't been touched - I only chanded the constructors for the arrays and replaced a header which contained some typedefs with a header containing this class. Do you have any idea what might be going on? Is there a better way to write an interface for successive [] operators? Thanks, Jiri

4 3

message_queue memory leak in Boost::test
by John Davies 09 Apr '13

09 Apr '13

I have this simple test case: BOOST_AUTO_TEST_CASE(TestLogin) { message_queue inQueue(open_or_create, "InQueue", 100, 100); remove("InQueue"); } And Boost::test reports a memory leak: Running 1 test case... *** No errors detected Detected memory leaks! Dumping objects -> {668} normal block at 0x0039DE78, 57 bytes long. Data: <struct boost::in> 73 74 72 75 63 74 20 62 6F 6F 73 74 3A 3A 69 6E {667} normal block at 0x0039DE40, 8 bytes long. Data: <x 9 > 78 DE 39 00 00 00 00 00 Object dump complete. Press any key to continue . . . If I comment out the two lines of code there is no memory leak. Is this something I have to live with? In my real code the message_queue is alive the whole time the program is running.

3 4

[Boost.Test + Turtle] Suggestion how to test async events.
by Daniele Barzotti 09 Apr '13

09 Apr '13

Hi, I'm trying to take a first touch with Boost.Test and Turtle. I have some objects for network communication with async notifications. class ServerHandler : public MessageEventHandler { public: ServerHandler() { // Register the messages I want to manage this->RegisterMsg(E_MSG_METHOD_CONNECT, Poco::delegate(this, &ServerHandler::OnMsgMethodConnect)); } // Messages Handlers virtual void OnMsgMethodConnect(const void* pSender, TArgs& args) { cout << "ServerHandler receive E_MSG_METHOD_CONNECT" << endl; MSG_METHOD_CONNECT& msg = RetrieveMessage<MSG_METHOD_CONNECT>(args.msg); }; }; This class is an observer which handle the notification through the OnMsgMethodConnect method. The test body I wrote is: // declare a 'mockServerHandler' class implementing 'ServerHandler' MOCK_BASE_CLASS( mockServerHandler, ServerHandler ) { MOCK_METHOD( OnMsgMethodConnect, 2 ); }; BOOST_AUTO_TEST_SUITE(NetMessageDispatcherTest) BOOST_AUTO_TEST_CASE(Test1) { mockServerHandler mockSvr; SessionFactory* sf = new SessionFactory( &mockSvr ); IMessageFactory* mf = new PBMessageFactory(); NetMessageDispatcher* server = new NetMessageDispatcher(sf, mf, NULL, E_DISP_SERVER); NetMessageDispatcher* client = new NetMessageDispatcher(sf, mf, NULL, E_DISP_CLIENT); BOOST_CHECK( server->Init("LOCALHOST", "TestServer") == 0 ); BOOST_CHECK( server->Connect(8000, "127.0.0.1") == 0 ); BOOST_CHECK( client->Init("LOCALHOST", "TestClient") == 0 ); BOOST_CHECK( client->Connect(8000, "127.0.0.1") == 0 ); MOCK_EXPECT( mockSvr.OnMsgMethodConnect ).once().with(????) } BOOST_AUTO_TEST_SUITE_END() Now, I have to test if into the ServerHandler::OnMsgMethodConnect() the argument args.msg (which is a message class) has the correct values. EG. if (msg.station_type() == EStationType::ST_CLIENT) => OK Which is the correct way to do it? Regards, Daniele.

2 2

file monitoring
by SRD 08 Apr '13

08 Apr '13

I'm looking for a way to get notified as soon as a new file has been created. So I was wondering if boost has a way of monitoring files so I can send them off as soon as I receive them. This would have to work for Linux and windows. I saw a few posts which seem to indicate that its not there yet, but that was some time ago. Thanks,

4 4

[boost::asio] how should I use accept and io_service?
by Shenyu Yao 08 Apr '13

08 Apr '13

Hi guys, this is my first post to this mail list. Forgive my noob questions, I am new to boost. I have read some documentations but still dont understand how io_service works. My problem is that I need two functions in my server class, listen() and handle() listen() is to accept all incoming connections and bind them with a handler function which accept a socket as argument. and handler will simply call io_service.run() ========================= class Server { io_service IOService; std::vector<boost:: ... :: socket> SocketVector; listen() { //blocking accept all incoming connection and bind socket with handleSocket } HandleRequest() { //io_service.run(); } } ============================ handleSocket(*Socket) { //handleSocket; } ============================== main() { boost::thread t1(&Server::listen, &myServer); boost::thread t2(&Server::HandleRequest, &myServer); } ==================================== my understanding is that everytime you instantiate a new socket object with a constructor including a io_service object, this socket object is associated with the io_service object, since I only have one io_service object, these socket object will be executed when I call io_service.run()but, the question is, how do i assign handle_function to those socket, I saw people use acceptor.async_accept(socket, boost::bind(handler, &socket)) like this, But I want to use accept rather than async_accept(), but the accept() function cant be used to bind a handler function to this socket Thank you

2 1

c++ Bit Fields
by Chris Gem 08 Apr '13

08 Apr '13

Please forgive me if this question has been answered but I've searched for days now and cannot find it. How do I expose a c++ struct with a bit field to python using boost::python? When I try I get this error. error: invalid pointer to bit-field ‘frameHdr_s::hdrSize’ Here is the struct: #define DATA_SIZE 0x100 typedef struct frameHdr_s { #define USE_BIT_FIELDS #ifdef USE_BIT_FIELDS // Longword 0 uint32_t hdrSize : 16; ///< Size of the header uint32_t revMin : 8; ///< Frame Minor Version uint32_t revMaj : 8; ///< Frame Major Version #else // Longword 0 uint32_t hdrSize; ///< Size of the header uint32_t revMin ; ///< Frame Minor Version uint32_t revMaj ; ///< Frame Major Version #endif } frameHdr_t; typedef struct frame_t { frameHdr_t frame_header; ///< frame header uint8_t data[DATA_SIZE]; ///< Byte array of data } frame_t; Here is boost code to expose it: using namespace boost::python; BOOST_PYTHON_MODULE(api) { class_<frameHdr_t>("frameHdr_t") .def_readwrite("hdrSize", &frameHdr_t::hdrSize) ; class_<frame_t>("frame_t") .def_readwrite("frame_header", &frame_t::frame_header) ; class_<myClass>("myClass") .def("register_callback", &myClass::register_callback) .def("boo", &myClass::boo) .def("crap", &myClass::crap) ; } I know that with ctypes I can create the same struct as in c++: from ctypes import * class frameHdr_t(Structure): _pack_ = 1 _fields_=[ # Longword 0 ("hdrSize",c_uint32, 16), ("revMin",c_uint32, 8), ("revMaj",c_uint32, 8), ] DATA_SIZE = 0x100 class frame_t(Structure): _pack_ = 1 _fields_=[ ("frame_header",frameHdr_t), ("data",c_uint8 * DATA_SIZE), ] But how do I translate from the Python version of frame_t to the c++ version of frame_t? In summary, * How do I tell boost to expose a bitfield? * If that is not possible, how do I translate from Python frame_t to C++ frame_t using boot? (I don't want to wrap in C because I need to preserve the C++ness of the interface.) The code is attached. Thank you for your time, Chris

4 4

type_erasure question
by Aleksandar Fabijanic 07 Apr '13

07 Apr '13

Looking at the type_erasure example: typedef any<boost::mpl::vector<copy_constructible<>, typeid_<>, addable<>> > any_type; any_type x(1.1); any_type y(1); any_type z(x + y); Apparently, without placeholders for x and y, this is undefined behavior. Am I interpreting it right? If so, shouldn't it fail to compile or at least throw an exception at runtime? Thanks, Alex

3 3

publish/subscribe/message architecture using boost
by Sean Farrow 07 Apr '13

07 Apr '13

Hi all, I 'm currently looking at redesigning an application and changing to boost/c++11 library components. The application reads data from an input, processes the data and then send it to an output. The current architecture is one thread per "device" where a device is a pair of input/output ports and one processing stage. What I'd like to do is move this to either a publish/subscribe model or some form of message-based system where an input comes in and a message is then sent to the next stage of the process. All stages will happen on the same physical/virtual machine. Are there any good pub/subscribe or message-based components using the boost architecture that works on mac osx/win32. Any help appreciated. Regards Sean.

3 3