- Boost-users - lists.preview.boost.org

Byte stream with multiple internal buffers
by Andrej van der Zee 16 Apr '11

16 Apr '11

Hi, I am writing a network sniffer that receives a variable number of TCP packets with payload strings of variable length that i need to concatenate. Using std string concatination is too expensive because of extra copying and possible reallocation of the internal buffer when it runs out of capacity. Better would be a special byte stream that stores its data in multiple variable-length buffers with an interface that works as if the buffers are just one buffer, with operations similar to std string. I was wondering if boost provides a supporting library for this. Thanks, Andrej

2 2

[karma] auto_ + default floating point precisions
by Brian Budge 15 Apr '11

15 Apr '11

Hi all - I'm using boost karma to generically generate strings from many different types use auto_. I've noticed that if I specify some double precision numbers like 1.123456, I will only get "1.123" out as a string. While looking around the documentation, it seems that it is possible to specify the precision up to numeric_limits<real>::digits10 + 1. I couldn't find an example for how to do this with double_ let alone auto_. Can anyone point me in the right direction? Is there documentation that google couldn't find for me? Thanks, Brian

2 2

[thread] : pause/resume thread
by Kazz 15 Apr '11

15 Apr '11

Hi All, I have just started learning boost. I want my thread to support following functions Pause(); Resume(); Stop();. I have come up with following code, which ends up as dead lock. Can some one please guide me as how to achieve the desired functionality. Would really appreciate any help. Thanks. Kaz class qthread { public: boost::thread* mThread; boost::condition_variable mCond; boost::mutex mMutex; bool mExit; qthread(): mExit(false) { // create function pointer to local Execute function boost::function<void (qthread*)> fptr = &qthread::Execute; fptr(this); // pass on function pointer along with this reference to boost thread boost::thread th0( fptr, this); // assign temp thread to local variable mThread = &th0; } void Resume() { mMutex.unlock(); mCond.notify_all(); } void Stop() { mThread->interrupt(); mExit = true; Resume(); } void Pause() { // todo } virtual void Execute() { std::cout<<"Entering thread Execution"<<std::endl; boost::unique_lock<boost::mutex> lock(mMutex); mCond.notify_all(); mCond.wait(lock); //<< DEAD LOCK int i = 0; while ( !mExit ) { try { while ( 1 ) { std::cout<<"Executing thread, cycle i = "<<++i<<std::endl; boost::this_thread::interruption_point(); } } catch( boost::thread_interrupted&) { std::cout<<"thread_interrupt exception"<<std::endl; mCond.wait(lock); } } std::cout<<"Exiting thread"<<std::endl; } }; //-------------------------- int main() { qthread qt; ///<< DEAD LOCK boost::this_thread::sleep(boost::posix_time::seconds(3) ); qt.Resume(); boost::this_thread::sleep(boost::posix_time::seconds(3) ); qt.Stop(); }

3 5

All-pairs SP algorithms -> Determine the path
by GOO Creations 15 Apr '11

15 Apr '11

Hi again I've noticed that the all-pairs SP algorithms (eg: Johnson, Floyd-Marshall) only "return" a distance matrix that hold the shortest distances between any node n and n'. Is there a way to determine the actual path (eg: n2-n4-n5-n1-n0) between these nodes using the all-pairs algorithms? Chris

3 2

Longest path from u to v in a directed weighted graph
by Shaun Jackman 15 Apr '11

15 Apr '11

Hi, I would like to find the longest path from vertex u to vertex v through a directed graph with positive edge weights. I know that u and v are choke points in the graph, in that if you start from vertex u and start following random edges, you will end up at vertex v in pretty short order. Is dijkstra_shortest_paths the best function for this purpose? The vertices reachable from u without travelling through v form a small subgraph of the total graph. I want to avoid exploring beyond v, and I definitely don't need to know the distances to vertices beyond v. Cheers, Shaun

4 6

Floyd-Warshall Example
by GOO Creations 15 Apr '11

15 Apr '11

Hi all I'm trying to figure out how to use the Floyd-Warshall algorithm. It seems like there is no example (like with the other algorithms); can someone maybe point me to an example. I'm having problems with the parameters, especially with the DistanceMatrix. Thank you Chris

2 2

[boost::iostreams] Is it possible to take char_type other than char or wchar_t
by 甜瓜 15 Apr '11

15 Apr '11

Hi, all, I have a proposal to make boost::iostreams support char_type other than char or wchar_t. Since the library gives users the opportunity to define char_type in Device, it should be available to boost::iostreams::copy from any source to any sink with the *same* char_type. Eg, If both Devices have UserDefinedStructure as char_type, boost::iostreams::copy should copy UserDefinedStructure objects from source to sink and return the number of objects copied. Currently, it is forbidden by a mistake (or not?) in boost::iostreams::copy. That is "detail::basic_buffer<char_type>" only allocate the raw space without construct them by using UserDefinedStructure constructor. I really appreciate this library, so would boost::iostreams please to concider my proposal? Or there are other ways to achieve this? Thank you in advance. -- leishen

1 0

boost.parameter - best practices
by Christoph Heindl 15 Apr '11

15 Apr '11

Hi, I have a set of classes that offer named parameter enabled constructors through the use of argument packs. Currently I have one huge header container all the available arguments of all classes. // params.h namespace geo { namespace keywords { BOOST_PARAMETER_NAME(a) BOOST_PARAMETER_NAME(b) } } // thing.h namespace geo { class thing { public: template<class Opts> thing(const Opts &o) { _m_x = o[Keywords::_a | 10]; //etc.. } }; } // thing2.h namespace geo { class thing2 { public: template<class Opts> thing2(const Opts &o) { _m_x = o[Keywords::_b | 10]; //etc.. } }; } First question: I'd like to place the BOOST_PARAMETER_NAME keywords as close to its usage as possible. I.e. I'd like to place BOOST_PARAMETER_NAME(a) on top of thing.h. If I do so, I think that over time other classes will start to have their own BOOST_PARAMETER_NAME(a) constructs in their header files, which I think would lead to multiple defined symbols if two class headers are included in the same translation unit, correct? So, is their any way to declare a parameter inside a class, through a typedef or similar? Let me illustrate class x { public: typedef boost::parameter::keyword<...> a; template<class Opts> x(const Opts &o) { _m_x = o[a | 10]; } }; x my_x(x::a = 5); If that is not possible, is there any way to limit the template parameter Opts in the above example to set of possible parameters? Best regards, Christoph

2 2

[Preprocessor] MSVC, File Iteration, and FLAGS()
by Robert Rumpf 14 Apr '11

14 Apr '11

I’ve reduced my problem to the following code snippet which compiles fine on GCC, but which fails to compile on MSVC (versions 9 and 10). The error provided by MSVC is “fatal error C1012: unmatched parenthesis : missing ')'”. It seems to have to do with the BOOST_PP_ITERATION_FLAGS() directive. I’m using Boost 1.45.0. Any thoughts on workarounds or bug-fixes? Thanks, Rob Rumpf *#if* !BOOST_PP_IS_ITERATING *# ifndef* __FILE_HPP__ *# define* __FILE_HPP__ *# include* <boost/preprocessor/iteration/iterate.hpp> *# include* <boost/preprocessor/seq/size.hpp> *# include* <boost/preprocessor/arithmetic/dec.hpp> /// Set up the first (most outer) nested FOR loop. *# define* INDICES_1 (A)(B) /// Set up the second (middle) nested FOR loop. *# define* INDICES_2 (C)(D) /// Set up the third (inner) nested FOR loop (1 of 3). *# define* INDICES_3a (E)(F) /// Set up the third (inner) nested FOR loop (2 of 3). *# define* INDICES_3b (G)(H) /// Set up the third (inner) nested FOR loop (3 of 3). *# define* INDICES_3c (I)(J) // Set up the number of iterations (see the Boost.Preprocessor library). // Set up the filename to be called (which is this one, see the Boost.Preprocessor library). *# define* BOOST_PP_ITERATION_PARAMS_1 (3, (0, BOOST_PP_DEC(BOOST_PP_SEQ_SIZE(INDICES_1)), "file.hpp")) // Iterate (see the Boost.Preprocessor library) on the first (most outer) nested FOR loop. *# include* BOOST_PP_ITERATE() *# endif* // __FILE_HPP__ // Choose the first iteration. *#elif* BOOST_PP_ITERATION_DEPTH() == 1 // Set up the number of iterations (see the Boost.Preprocessor library). // Set up the filename to be called (which is this one, see the Boost.Preprocessor library). *# define* BOOST_PP_ITERATION_PARAMS_2 (3, (0, BOOST_PP_DEC(BOOST_PP_SEQ_SIZE(INDICES_2)), "file.hpp")) // Iterate (see the Boost.Preprocessor library) on the second (middle) nested FOR loop. *# include* BOOST_PP_ITERATE() // Choose the second iteration. *#elif* BOOST_PP_ITERATION_DEPTH() == 2 // Set up the number of iterations (see the Boost.Preprocessor library). // Set up the filename to be called (which is this one, see the Boost.Preprocessor library). *# define* BOOST_PP_ITERATION_PARAMS_3 (4, (0, BOOST_PP_DEC(BOOST_PP_SEQ_SIZE(INDICES_3a)), "file.hpp", 0)) // Iterate (see the Boost.Preprocessor library) on the third (inner) nested FOR loop (1 of 3). *# include* BOOST_PP_ITERATE() // Set up the number of iterations (see the Boost.Preprocessor library). // Set up the filename to be called (which is this one, see the Boost.Preprocessor library). *# define* BOOST_PP_ITERATION_PARAMS_3 (4, (0, BOOST_PP_DEC(BOOST_PP_SEQ_SIZE(INDICES_3b)), "file.hpp", 1)) // Iterate (see the Boost.Preprocessor library) on the third (inner) nested FOR loop (2 of 3). *# include* BOOST_PP_ITERATE() // Set up the number of iterations (see the Boost.Preprocessor library). // Set up the filename to be called (which is this one, see the Boost.Preprocessor library). *# define* BOOST_PP_ITERATION_PARAMS_3 (4, (0, BOOST_PP_DEC(BOOST_PP_SEQ_SIZE(INDICES_3c)), "file.hpp", 2)) // Iterate (see the Boost.Preprocessor library) on the third (inner) nested FOR loop (3 of 3). *# include* BOOST_PP_ITERATE() // Look for the third nested loop, but index over INDICES_3a this time. *#elif* BOOST_PP_ITERATION_DEPTH() == 3 && BOOST_PP_ITERATION_FLAGS() == 0 // Look for the third nested loop, but index over INDICES_3b this time. *#elif* BOOST_PP_ITERATION_DEPTH() == 3 && BOOST_PP_ITERATION_FLAGS() == 1 // Look for the third nested loop, but index over INDICES_3c this time. *#elif* BOOST_PP_ITERATION_DEPTH() == 3 && BOOST_PP_ITERATION_FLAGS() == 2 *#endif* // BOOST_PP_IS_ITERATING

1 0

[boost::threads] -D_REENTRANT?
by Panagiotis Foteinos 14 Apr '11

14 Apr '11

Hello users. I am using boost::threads for an application that allocates space in heap with the new operator. Therefore, different threads allocate memory in parallel. Do I have to link my application to any other library apart from the libboost_thread to guarantee a thread safe memory allocation? I searched online, but I got more confused. Do I have to define the -D_REENTRANT or the -pthread symbol? Best Regards, Panagiotis Foteinos

3 3