- Boost-users - lists.boost.org

[Iostreams] Problem to chain gzip filter with home-made one
by Christophe hugon 31 Oct '13

31 Oct '13

Hello everybody, I meeting a execution issue while using the gzip filter of the Boost.Iostreams, combined with a buffered home-made one. I would like to know what is wrong in my way to do it. Firstly, my filter is a general DualFilter. It uses an buffer (of 16 bytes) before the transformation and the put/return. The categories of my filter is closable and dual_use. You can find code details in the end of this mail. My problem is ([a/b] means for the separates case with the letter a and b): If I take the 6 possibilities for write and read: 1) io::filtering_[i/o]stream stm; 2) stm.push(io::gzip_[de/]compressor()); 3) stm.push(my_[Unt/T]ransFilter); at the moment to use the stm.[get/put] to read/write 1)-2) works perfectly 1)-3) works perfectly 1)-3)-2) works perfectly 1)-2)-3) Stay stuck (obviously it is the one I really need) In fact in the last case, the gzip filter give this infinite loop to the put of the TransFilter : 31 -117 8 0 0 0 0 0 0 31 -117 8 0 0... A similar case occur with the bzip filter (even if it is a bit longer loop). What do I do wrong ? Somebody can help me ? Thanks in advance Christophe PS: here a simplified implementation of my filter template<typename Source> int get(Source& src) { //at the very beginning or if the buffer are fully returned if (fcurrent_pos == fblkLength) { fcurrent_pos =0; FillThem (src); //fill&transform the buffers with the right number of src.get() fc=CheckNSolveBufferIsEnd (src); //eventually prepare read the suffix of the file } //The effective return if (fcurrent_pos < fUntrans.size ()) return (int)fUntrans[fcurrent_pos++]; else return fc; } template<typename Sink> bool put(Sink& dest, int c) { if (c==EOF || c ==boost::iostreams::WOULD_BLOCK) return false; if (fUntrans.size () < fblkLength - 1) { fUntrans.AddData((char*)(&c),1); return true; } fUntrans.AddData((char*)(&c),1); fenc->TransformThis (fUntrans, fTrans, fKey); for (int i=0; i<fblkLength; i++) { if (!boost::iostreams::put(dest, (int)fTrans[i])) return false; } return true; } template<typename Sink> bool close (Sink &dest, BOOST_IOS::openmode m) { if (fFlushed) return true; PrepareSuffix (); for (int i=0; i< fblkLength*2; //suffix length i++) { if (!boost::iostreams::put(dest, (int)fTrans[i])) return false; } return true; }

1 0

BOOST ASIO: Having successful READ from the SERVER ..??
by Rahul Mathur 30 Oct '13

30 Oct '13

All, Initially after having done successful CONNECT with the server, sending successful LOGON message and finally KEEP ALIVE status, I tried to read from the server as - -- boost::asio::streambuf data; std::size_t length = boost::asio::read(socket, data, boost::asio::transfer_at_least(socket.available())); -- but was able to partially receive response from remote server which I don't think was completely successful, so to have COMPLETE RESPONSE from the server, tried with below modification - -- boost::system::error_code ec; boost::asio::streambuf data; std::size_t length = boost::asio::read(socket, data, boost::asio::transfer_all(), ec); --- but this modification doesn't work. Any clue what should be the proper way? Thanks!!

2 1

SIGSEGV when removing Ethernet/Serial Port cables
by Ulf Samuelsson 30 Oct '13

30 Oct '13

Have a boost enable application which communicated over USB serial ports and UDP If the cables are removed, the library fires off SIGSEGV signals. I am trying to catch exceptions but neither } catch (boost::system::system_error &e) { } catch (std::exception &e) { seems to work, and a SIGSEGV signal is sent when the port->get_io_service().reset(); statement is executed. How can I handle the problem, that the USB serial cable is removed? I know I can test for the existence of /dev/ttyUSBx in the file system or /sys/class/tty/ttyUSBx" I know I can write a SIGSEGV handler, but would like to know if there is any exception I can catch instead. Have the same problem when the Ethernet Cable is removed. How handle this? // Serial Port Read. // Reads a character or times out bool SerialPort::read_char(unsigned char& val) { val = c = '\0'; try { // After a timeout & cancel it seems we need // to do a reset for subsequent reads to work. port->get_io_service().reset(); // Asynchronously read 1 character. async_read(*port, buffer(&c, 1), boost::bind(&SerialPort::read_complete, this, placeholders::error, placeholders::bytes_transferred)); // Setup a deadline time to implement our timeout. timer->expires_from_now(boost::posix_time::milliseconds(1)); timer->async_wait(boost::bind(&SerialPort::time_out, this, placeholders::error)); // This will block until a character is read // or until the it is cancelled. port->get_io_service().run(); if (!read_error) { val = c; } else { val = 0; // Timeout } } catch (boost::system::system_error &e) { val = 2; DebugPort.WriteLine("Serial Port System Error", RED); return false; } catch (std::exception &e) { val = 1; DebugPort.WriteLine("Serial Port General Error", RED); return false; } return !read_error; } // Network port open bool UdpClient::Open(void) { udp::resolver resolver(io_service); udp::resolver::query query(udp::v4(), ipAddress, Port); try { Endpoint = *resolver.resolve(query); } catch (boost::system::system_error &e) { // We might not have a network connection return false; } if (!CreateSocket()) { return false; } if (!OpenSocket()) { return false; } init = true; return true; } Best Regards Ulf Samuelsson

2 1

Looking for official .m4 script
by Yaroslav Panych 30 Oct '13

30 Oct '13

HI Looking for official .m4 script to include it into distribution package in order to use its benefits while configuring/build software. Yes, I know about Tsuna's boost.m4 library but I never heard it was approved by Boost officials. Regards

2 1

[graph] ambiguity moving from 1.48 to 1.54
by Richi Lists 30 Oct '13

30 Oct '13

Hi, I'm in the process of upgrading from boost 1.48 to 1.54. We're using MSVC10. Most stuff that comes up is easy to fix, but this one gives me headaches: ...boost\boost_1_54_0\boost/pending/property.hpp(108): error C2752: 'boost::lookup_one_property_internal<PList,PropName>' : more than one partial specialization matches the template argument list with [ PList=boost::property<boost::edge_bundle_t,EdgeData>, PropName=boost::edge_bundle_t ] ...boost\boost_1_54_0\boost/pending/property.hpp(114): could be 'boost::lookup_one_property_internal<boost::property<Tag,T,Base>,Tag>' ...boost\boost_1_54_0\boost/pending/property.hpp(108): or 'boost::lookup_one_property_internal<boost::property<Tag,T,Base>,boost::edge_bundle_t>' ...boost\boost_1_54_0\boost/pending/property.hpp(108): or 'boost::lookup_one_property_internal<T,boost::edge_bundle_t>' ...boost\boost_1_54_0\boost/pending/property.hpp(158) : see reference to class template instantiation 'boost::lookup_one_property_internal<PList,PropName>' being compiled with [ PList=boost::property<boost::edge_index_t,int,boost::property<boost::edge_bundle_t,EdgeData>>, PropName=boost::edge_bundle_t ] ...boost\boost_1_54_0\boost/graph/detail/adjacency_list.hpp(1505) : see reference to class template instantiation 'boost::lookup_one_property<T,Tag>' being compiled with [ T=boost::property<boost::edge_index_t,int,boost::property<boost::edge_bundle_t,EdgeData>>, Tag=boost::edge_bundle_t ] ...boost\boost_1_54_0\boost/graph/detail/adjacency_list.hpp(2028) : see reference to class template instantiation 'boost::adj_list_helper<Config,Base>' being compiled with [ Config=boost::detail::adj_list_gen<boost::adjacency_list<boost::vecS,boost::vecS,boost::undirectedS,VertexProps,EdgeProps>,boost::vecS,boost::vecS,boost::undirectedS,VertexProps,EdgeProps,boost::no_property,boost::listS>::config, Base=boost::undirected_graph_helper<boost::detail::adj_list_gen<boost::adjacency_list<boost::vecS,boost::vecS,boost::undirectedS,VertexProps,EdgeProps>,boost::vecS,boost::vecS,boost::undirectedS,VertexProps,EdgeProps,boost::no_property,boost::listS>::config> ] ...boost\boost_1_54_0\boost/graph/adjacency_list.hpp(267) : see reference to class template instantiation 'boost::vec_adj_list_impl<Graph,Config,Base>' being compiled with [ Graph=boost::adjacency_list<boost::vecS,boost::vecS,boost::undirectedS,VertexProps,EdgeProps>, Config=boost::detail::adj_list_gen<boost::adjacency_list<boost::vecS,boost::vecS,boost::undirectedS,VertexProps,EdgeProps>,boost::vecS,boost::vecS,boost::undirectedS,VertexProps,EdgeProps,boost::no_property,boost::listS>::config, Base=boost::undirected_graph_helper<boost::detail::adj_list_gen<boost::adjacency_list<boost::vecS,boost::vecS,boost::undirectedS,VertexProps,EdgeProps>,boost::vecS,boost::vecS,boost::undirectedS,VertexProps,EdgeProps,boost::no_property,boost::listS>::config> ] ..\graph_problem.cpp(52) : see reference to class template instantiation 'boost::adjacency_list<OutEdgeListS,VertexListS,DirectedS,VertexProperty,EdgeProperty>' being compiled with [ OutEdgeListS=boost::vecS, VertexListS=boost::vecS, DirectedS=boost::undirectedS, VertexProperty=VertexProps, EdgeProperty=EdgeProps ] ...boost\boost_1_54_0\boost/graph/adjacency_list.hpp(374): error C2182: '[]' : illegal use of type 'void' The code to reproduce is here: // boost #include <boost/graph/adjacency_list.hpp> #include <boost/graph/subgraph.hpp> #include <boost/graph/graph_traits.hpp> class EDGE {}; ////////////////////////////////////////////////////////////////////////// /// graph's vertex data struct VertexData { /// c-tor VertexData() { pos[0] = 0.0; pos[1] = 0.0; pos[2] = 0.0; } VertexData(const double* _pos) { pos[0] = _pos[0]; pos[1] = _pos[1]; pos[2] = _pos[2]; } /// data double pos[3]; }; ////////////////////////////////////////////////////////////////////////// /// graph's edge data struct EdgeData { /// c-tor EdgeData() : pEdge(NULL) {} EdgeData(EDGE* _pEdge) : pEdge(_pEdge) {} /// comparison for equality bool operator==(const EdgeData& dst) const { return pEdge == dst.pEdge; } /// data EDGE* pEdge; }; ////////////////////////////////////////////////////////////////////////// /// types typedef VertexData VertexProps; ///< custom vertex properties typedef boost::property<boost::edge_index_t, int, boost::property<boost::edge_bundle_t, EdgeData> > EdgeProps; ///< custom edge properties typedef boost::adjacency_list<boost::vecS, boost::vecS, boost::undirectedS, VertexProps, EdgeProps> Type; ///< custom graph type ////////////////////////////////////////////////////////////////////////// /// unary predicate for testing if edge has duplicates in graph struct CheckDuplicateEdge { CheckDuplicateEdge(const Type& _g) : g(_g) {} // check if this edge has duplicates bool operator()(const Type::edge_descriptor& e) const { // do something return false; } private: const Type& g; }; ////////////////////////////////////////////////////////////////////////// void RemoveDuplicateEdges(Type& g) { remove_edge_if(CheckDuplicateEdge(g), g); } ////////////////////////////////////////////////////////////////////////// void main() { std::cout << "hello graph"; } ////////////////////////////////////////////////////////////////////////// I have tried to find a solution, and have searched the internet for hours. So, hopefully anybody here saw something similar, and solved it already. Rgds Richard

2 3

[python] Crash right after initialization after ubgrading from 1.48 to 1.54
by Richi Lists 30 Oct '13

30 Oct '13

I upgraded from boost 1.48 to .154 and I thought I was done after all the compiler errors and unit tests were ironed out. But now when I start our application, it terminates immediately. This only happens with the debug build. The release build is fine. I verified that it properly links to boost-python-vc100-mt-gyd-1_54.lib and python27_d.lib The python version (2.73) is still the same that we used with boost 1.48. I tracked it down to here: Py_Initialize(); m_mainModule = boost::python::import(__main__); <-- crash here With the debugger I found out that the string object is invalid. More specifically, ob_type is 0xffffffff. So I added this exception which always throws now: Py_Initialize(); const boost::pathon::str someRandomString("someRandomString"); if(reinterpret_cast<void*>(0xffffffff) == someRandomString.ptr()->ob_type) throw std::runtime_error("Python problem"); What could be going on here? Rgds Richard

1 0

[Geometry] Compilation errors after moving to 1.55b1
by Szymon Gatner 29 Oct '13

29 Oct '13

Moving to 1.55b1 from 1.54 gives: Error 1 error C2976: 'boost::geometry::strategy::transform::translate_transformer' : too few template arguments offending line is: typedef strategy::transform::translate_transformer<Point, Point> Translate; Regards, Szymon Gatner

3 3

[Asio] Compilation errors after switching to 1.55 b1
by Szymon Gatner 29 Oct '13

29 Oct '13

I am getting very weird error after moving from 1.54 to 1.55b1: 2>C:\devel\boost_1_55_0b1\boost/asio/detail/impl/socket_ops.ipp(2224): error C2653: 'Windows' : is not a class or namespace name (..\..\lib\hexine.cpp) and many more in that file, then: 2>C:\devel\boost_1_55_0b1\boost/asio/detail/buffer_sequence_adapter.hpp(36): error C2653: 'Windows' : is not a class or namespace name (..\..\lib\hexine.cpp) Than, many errors later: 2>C:\devel\boost_1_55_0b1\boost/asio/detail/winrt_async_manager.hpp(62): error C2039: 'IAsyncAction' : is not a member of 'Windows::Foundation' (..\..\lib\hexine.cpp) This looks like C++/CLI implementation. I assume this is related to "Implemented a limited port to Windows Runtime. This support requires that the language extensions be enabled." Does it mean that switch from 1.54 to 1.55 requires changes in project settings? Visual Studio 2012 Update 3, Windows 7 Regards, Szymon Gatner

2 1

[date_time] Not built with Visual Studio 2013
by Szymon Gatner 29 Oct '13

29 Oct '13

Hi, I just built boost 1.55 beta1 with recently released VC2013 on Win7 and for some reason date-time library was not built. I didn't sepcify it when building (just hit b2 to get everything). Regards, Szymon Gatner

2 3

[Boost-users] {exception][boost::throw_exception][BOOST_NO_EXCEPTIONS] implications
by Jeff Flinn 28 Oct '13

28 Oct '13

The project I've joined [ab]uses BOOST_NO_EXCEPTIONS in 1.55b as a means to intercept exceptions thrown via boost::throw_exception() and translate them to the project's single exception type derived from std::exception, leaving me a bit queasy. That queasiness is reinforced by #ifdef BOOST_NO_EXCEPTIONS #error This header requires exception handling to be enabled. #endif in ...\boost_1_55_0b1\boost\exception\detail\exception_ptr.hpp. This implies that BOOST_NO_EXCEPTIONS is used for more that just allowing client code to supply it's own throw_exception. Does boost and/or C++ provide a better customization point to safely translate exceptions thrown by boost? (note: we're using vs2013, generally the latest gcc, and clang) Thanks, Jeff

2 1