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

Re: [Boost-users] Boost.Asio -- combining sync read with async read
by Sameer Parekh 14 Oct '09

14 Oct '09

> > No, it will read anything available in the socket and put it in the > buffer. > It can be 1 line, 2 lines or 3.25 lines. What I mean is that you > have to be > aware of the fact that the buffer might contain more data than you > "requested" from read_until. Aha, thanks Igor. So my handler needs to be able to deal with a buffer that contains more than one line. That's not so bad, I can just run the processing code for each line until the buffer is empty, that should be fine. However, what concerns me is if the handler receives a buffer that has just 2.5 lines in it. All the messages from the server are going to be one line long. So if the server sends three messages in succession but say there is some network hiccup partway through the 3rd line, then my client will only receive 2.5 lines, and the handler will get called with the partial line? Then the handler has to cache the partial line so that when the rest of the line comes through it can paste them together again? That seems hard... thanks, -s

2 1

[asio] deadline_timer callback is not called
by Daniele Barzotti 14 Oct '09

14 Oct '09

Hi, I've yet posted a question about this topic but I not solve, so I try to ask again. PS: I used the same solution (in conjunction with asio::serial_port) in another project and, there, it works! I'm working on a library and I have only one static io_service and io_service::work like in the following code: ---------------------- START CODE ------------------------ typedef boost::shared_ptr<boost::asio::io_service::work> io_work_ptr; // Main io_service static boost::asio::io_service io_service; static boost::scoped_ptr<boost::thread> io_service_thread; static io_work_ptr p_work; static bool thread_started; void io_worker_thread(void) { #if(WIN32 && _WIN32_DCOM) struct com_init { com_init() { CoInitializeEx(NULL, COINIT_APARTMENTTHREADED); } ~com_init() { CoUninitialize(); } } initializer_object; #endif io_service.run(); }; bool CreateMainThread() { if (!thread_started) { try { // create the work object on the heap: // [Reference counting] // This object is a simple pointer because // it is assigned to each created objects // These objects store it in a shared_ptr so, // when the last object is destroyed // The io_service::work is destroyed as well p_work.reset( new boost::asio::io_service::work(io_service) ); // Instead of start the worker thread directly by // passing the io_service object, // I call the io_worker_thread function where // I can do some initialization tasks like // calling CoInitializeEx for COM io_service_thread.reset(new boost::thread(io_worker_thread)); thread_started = !thread_started; return true; } catch (boost::thread_resource_error e) { // Failed to create the new thread return false; } } else return true; } /*! * \brief Factory function for AudioPlayer objects creation */ EXTERN_C EUROAUDIOLIB_API AudioPlayerHndl CALL CreateAudioPlayer() { if (!CreateMainThread()) return NULL; // create the new object AudioPlayer* pAudioPlayer = new AudioPlayer; pAudioPlayer->io_worker_ptr = p_work; return pAudioPlayer; }; ---------------------- END CODE ------------------------ In the above code pAudioPlayer->io_worker_ptr is declared as: io_work_ptr p_work; In a class I have to use the timer, so I initialize it with the global io_service object: ---------------------- START CODE ------------------------ boost::asio::deadline_timer rx_timer_; WaveStream::WaveStream() : rx_timer_(io_service) { }; void WaveStream::ProcessRTPPacket() { // RTP Packet received ...... //Start a new timer or renew it rx_timer_.expires_from_now( boost::posix_time::milliseconds( 1000 ) ); io_service.reset(); // We managed to cancel the timer. Start new asynchronous wait. rx_timer_.async_wait( boost::bind(&WaveStream::handle_timeout, this, _1) ); .... } void WaveStream::handle_timeout(const boost::system::error_code& error) { std::cout << "BOOST TIMER RAISED!!" << endl; // A Timeout is raised: a stream is end if (!error) {...} }; ---------------------- END CODE ------------------------ But the callbak is never called!! Since ProcessRTPPacket is called frequently, I try also to move the timer on another point just for test with the same effect! Where could be the mistake? Thanks in advance! Daniele.

2 11

Re: [Boost-users] Boost.Asio -- combining sync read with async read (Igor R)
by Sameer Parekh 14 Oct '09

14 Oct '09

> > It's ok to use both sync. and async. calls, but you should not call > boost::asio::async_read_until() and/or boost::asio::async_read() until > previous read is complete (for the same socket/buffer). Besides, pay > attention that reading "until" might read more data than you expect. Thanks Igor, I'm a bit confused here though.. how do I know if the previous read is complete? I figured when async_read_until() calls the handler, the read was complete and everything read was sitting in the buffer? Do you mean to say that if the client receives two lines, then async_read_until() will read both lines but then put only the first line in the buffer? Then I can't call async_read_until() again because there is still another pending call to the handler that will process the second line? So how do I know at the end of the handler whether or not to call async_read_until() ? thanks! -Sameer

2 1

[fusion] probably a bug in result_of::join
by jl 14 Oct '09

14 Oct '09

This is nearly a copy-paste from the doc (and agrees with intuition): #include <boost/fusion/algorithm/transformation/join.hpp> #include <boost/fusion/include/vector.hpp> #include <boost/fusion/include/join.hpp> void foo() { using namespace boost::fusion; vector<> v1, v2; result_of::join< vector<>, vector<> >::type j = join(v1, v2); } ...but it doesn't compile (msvc9, boost 1.40): c:\UserTemp\x54209\BoostTests\Any\main.cpp(20) : error C2440: 'initializing' : cannot convert from 'boost::fusion::result_of::join<LhSequence,RhSequence>::type' to 'boost::fusion::joint_view<Sequence1,Sequence2>' with [ LhSequence=const boost::fusion::vector<>, RhSequence=const boost::fusion::vector<> ] and [ Sequence1=boost::fusion::vector<>, Sequence2=boost::fusion::vector<> ] No constructor could take the source type, or constructor overload resolution was ambiguous J-L

2 3

[Variant] Sorting data out of a vector of variant
by Oliver Rudolph 14 Oct '09

14 Oct '09

Hi all! I have some class with a member which is a vector of boost.variant, i.e. class Example { public: std::vector<variant<typeA, typeB, typeC> > content; } I want to add three members to my class that are vectors of pointers to the specific content items class Example { public: std::vector<variant<typeA, typeB, typeC> > content; std::vector<typeA> *contentA; std::vector<typeB> *contentB; std::vector<typeC> *contentC; } Now the question: how can I iterate through the vector content whilst adding its items to the other 3 vectors depending on their type? I know there has to be some way using the apply_visitor functor, but I just don't see how exactly... Thanks for your help! Best regards, Oli

2 1

Re: [Boost-users] Boost.Serialization not building
by OvermindDL1 14 Oct '09

14 Oct '09

On Wed, Oct 21, 2009 at 8:53 AM, Stefan Strasser <strasser(a)uni-bremen.de> wrote: > >> >> So what do I need to do to the xml archive to get it to work like the >> others? _______________________________________________ > > use NVPs I guess > not sure if that's the cause of THIS error though I tried looking at that, but have no clue what that is (other then some lines of code referencing it I found), or how I could apply it to this situation. Should it not work the same for the xml archive as it does text and binary, which both work fine.

2 1

Boost.Asio -- combining sync read with async read
by Sameer Parekh 14 Oct '09

14 Oct '09

Hello everyone! I am writing a client for a protocol which goes something like this: The client connects to the server, and it submits "orders" -- every time it gets an order it gets back an "ack". However, the server might asynchronously send a "fill" at any time. So I want to be listening for a fill even while the client might be doing other things (listening to another server for one-way data stream. The client then chooses to submit an order or not based on the information it gets in the datastream, or it might choose to submit an order if it receives a fill.) I thought the way to do it would be to do async_read_until() and have it call my fill handler when it receives something. Then the function that submits orders would call write() and then block waiting for the ack message using read_until() However, when I implement that, the client never seems to notice that a fill message is sent. It receives the acks just fine, and it successfully reads the datafeed coming in from the other tcp connection. I thought maybe the read_until() would cancel the async_read_until() so after I received the ack message I called async_read_until() again, but that didn't work either... (The server works as it should, I tested it with telnet.) Is it possible to mix sync/async reads in this way? Can anyone tell if I am doing something wrong? Do I need to do something else to make this work? Thanks, I appreciate your help! Here is the code in question: (as an aside, is it bad form to call connect() in the constructor? Should I call connect() and setup the async_read_until() in a separate start() method?) MarketConnectionTCP::MarketConnectionTCP(boost::asio::io_service &io_service_, std::string ip_, unsigned short agentPort_, unsigned short feedPort_) : _io_service(io_service_), _agentSocket(_io_service), _feedSocket(_io_service), _timer(_io_service, boost::posix_time::seconds(1)) { boost::asio::ip::address ipAddress(boost::asio::ip::address::from_string( ip_)); tcp::endpoint agentEndpoint(ipAddress, agentPort_); Logger::stream() << "Trying to connect to agent:" << agentEndpoint << std::endl; _agentSocket.connect(agentEndpoint); Logger::stream() << "Agent connected:" << _agentSocket.remote_endpoint() << std::endl; boost::asio::ip::tcp::endpoint feedEndpoint(ipAddress, feedPort_); Logger::stream() << "Trying to connect to feed:" << feedEndpoint << std::endl; _feedSocket.connect(feedEndpoint); Logger::stream() << "Feed connected:" << _feedSocket.remote_endpoint() << std::endl; boost::asio::async_read_until(_agentSocket, _fillBuf, "\r\n", boost::bind( &MarketConnectionTCP::handleReadFill, this, boost::asio::placeholders::error)); Logger::stream() << "Listening on agent connection:" << _agentSocket.remote_endpoint() << std::endl; boost::asio::async_read_until(_feedSocket, _feedBuf, "\r\n", boost::bind( &MarketConnectionTCP::handleReadFeed, this, boost::asio::placeholders::error)); Logger::stream() << "Listening on feed connection:" << _feedSocket.remote_endpoint() << std::endl; _timer.async_wait(boost::bind(&MarketConnectionTCP::handleTimer, this, boost::asio::placeholders::error)); Logger::stream() << "Timer is waiting" << std::endl; } OrderID MarketConnectionTCP::sendOrder(const Order &order) { boost::asio::streambuf orderBuf; std::ostream orderStream(&orderBuf); orderStream << order << "\r\n"; // This blocks until we get an ack with the order id Logger::stream() << "Sending order to:" << _agentSocket.remote_endpoint() << ":" << order << std::endl; boost::asio::write(_agentSocket, orderBuf); Logger::stream() << "Waiting for ack from:" << _agentSocket.remote_endpoint() << std::endl; boost::asio::streambuf ackBuf; std::istream ackStream(&ackBuf); boost::asio::read_until(_agentSocket, ackBuf, "\r\n"); OrderID id; ackStream >> id; Logger::stream() << "Received ack from:" << _agentSocket.remote_endpoint() << ":" << id << std::endl; // listen again boost::asio::async_read_until(_agentSocket, _fillBuf, "\r\n", boost::bind(&MarketConnectionTCP::handleReadFill, this, boost::asio::placeholders::error)); Logger::stream() << "Listening on agent connection:" << _agentSocket.remote_endpoint() << std::endl; return id; } void MarketConnectionTCP::handleReadFill(const boost::system::error_code& err) { if (!err) { std::istream fillStream(&_fillBuf); Fill newFill; fillStream >> newFill; if (newFill.isValid()) { Logger::stream() << "Fill received on:" << _agentSocket.remote_endpoint() << ":" << newFill << std::endl; receiveFill(newFill); } else { Logger::stream() << "Invalid fill received on:" << _agentSocket.remote_endpoint() << std::endl; } // listen again boost::asio::async_read_until(_agentSocket, _fillBuf, "\r\n", boost::bind(&MarketConnectionTCP::handleReadFill, this, boost::asio::placeholders::error)); Logger::stream() << "Listening on agent connection:" << _agentSocket.remote_endpoint() << std::endl; } else { delete this; } }

2 1

Boost compatibility with Old Compilers
by chin 14 Oct '09

14 Oct '09

HI, We are planning to use Boost version 1.39 on Windows. In docs i found the following: ---------------------------------------------------------------------------------------------------- Compilers Tested Boost's primary test compilers are: o Linux: o GCC 4.3.3 on Ubuntu Linux. o Windows: o Visual C++ 7.1 SP1, 8.0 SP1 and 9.0 SP1 on Windows XP. Boost's additional test compilers include: o Linux: o Intel 9.0 on Red Hat Enterprise Linux. o Intel 10.0 on Red Hat Enterprise Linux. o Intel 10.1 on 64-bit Linux Redhat 5.1 Server. o Intel 10.1 on Suse Linux on 64 bit Itanium. o Intel 11.0 on Red Hat Enterprise Linux. o Intel 11.1 on Red Hat Enterprise Linux. o GCC 3.4.3, GCC 4.0.1, GCC 4.2.4, GCC 4.3.3 and GCC 4.4.0 on Red Hat Enterprise Linux. o GCC 4.3.3 and GCC 4.4.0 with C++0x extensions on Red Hat Enterprise Linux. o GCC 4.1.1, 4.2.1 on 64-bit Red Hat Enterprise Linux. o GCC 4.1.2 on Suse Linux on 64 bit Itanium. o GCC 4.1.2 on 64-bit Redhat Server 5.1. o GCC Open64 4.2.2 on Red Hat Enterprise Linux. o GCC 4.3.4 on Debian unstable. o QLogic PathScale(TM) Compiler Suite: Version 3.1 on Red Hat Enterprise Linux. o GCC version 4.2.0 (PathScale 3.2 driver) on 64-bit Red Hat Enterprise Linux. o Sun 5.9 on Red Hat Enterprise Linux. o Windows: o Visual C++ 7.1 on XP. o Visual C++ 9.0 on XP. o Visual C++ 9.0 on Vista. o Visual C++ 9.0 on Vista 64-bit. o Visual C++ 9.0, using STLport 5.2, on XP and Windows Mobile 5.0. o Visual C++ 10.0 beta. o Borland 5.9.3, 6.1.0, 6.1.3. o Borland C++ Builder 2007 and 2009. o Intel C++ 11.1, with a Visual C++ 9.0 backend, on Vista 32-bit. o Intel C++ 11.1, with a Visual C++ 9.0 backend, on Vista 64-bit. o GCC 4.3.3 and 4.4.0, on Mingw with C++0x features. o Solaris: o Sun C++ 5.7, 5.8, 5.9 on Solaris 5.10. o GCC 3.4.6 on Solaris 5.10. ---------------------------------------------------------------------------------------------------- My concern is: We are supposed to use Boost both for Linux and Solaris, but the compilers being used are slightly old. On Debian Linux we have GCC 3.3 and on Solaris we have Sun WorkShop 6 update 2 C++ 5.3. Is Boost compatible with these versions? Does this list mean that Boost is not compatible with older versions of Compiler? -- View this message in context: http://www.nabble.com/Boost-compatibility-with-Old-Compilers-tp25885255p258… Sent from the Boost - Users mailing list archive at Nabble.com.

2 1

TCP boost server, but non-boost client
by G, Benjamin (NSN - IN/Bangalore) 14 Oct '09

14 Oct '09

Hi, I have a server which provides configuration to different clients. This server uses boost's serialization to serialize the C++ configuration objects. Some of my clients don't want to use boost for the (de-)serialization work. Will this cause any issues? Can any client deserialize the data serialized by a boost-using entity? What are the things that I need to take care of? I did read somewhere that boost serialization should be portable (I wonder if it means the same). All help appreciated. Regards, - Ben

2 1

[asio] How to know if the io_service is runing
by Daniele Barzotti 14 Oct '09

14 Oct '09

In my library I have this code (yet used in another project) where pwork is assigned to a shared_ptr: -------------------------------------------------------------- typedef boost::shared_ptr<boost::asio::io_service::work> io_work_ptr; // Main io_service static boost::asio::io_service io_service; static boost::scoped_ptr<boost::thread> io_service_thread; static boost::asio::io_service::work* p_work; static bool thread_started; bool CreateMainThread() { if (!thread_started) { try { // create the work object on the heap p_work = new boost::asio::io_service::work(io_service); // run the IO service as a separate thread: io_service_thread.reset ( new boost::thread ( boost::bind ( &boost::asio::io_service::run, &io_service ) ) ); thread_started = !thread_started; return true; } catch (boost::thread_resource_error e) { // Failed to create the new thread return false; } } } -------------------------------------------------------------- WaveStream::WaveStream() : rx_timer_(io_service) { }; WaveStream::Test() { //Start a new timer or renew it rx_timer_.expires_from_now( boost::posix_time::milliseconds(500) ); // We managed to cancel the timer. Start new asynchronous wait. rx_timer_.async_wait( boost::bind(&WaveStream::handle_timeout, this, _1) ); } void WaveStream::handle_timeout(const boost::system::error_code& error) { .... } -------------------------------------------------------------- The timer callback is never called. Is there a way to know if the io_service is runing? Daniele.

2 2