September 2014 - Boost-users

[Threads] and OpenMP
by Stephan Menzel 01 Oct '14

01 Oct '14

Hi all, I would like to ask around about a strategy on how to tacke the following threading problem: I am maintaining a boost intensive portable C++ client-server application and recently have included a highly complex 3rd party algorithm into my application that does CPU intense stuff. Long running stuff. Often in the range of days. While at the same time being heavy in OpenMP usage. Now of course requirements are that the user has to be able to cancel this. The algorithm runs in a boost threads controlled background thread. Now I began to place interruption points throughout the 3rd party code but soon had to realize that a lot of CPU time is being handled in OpenMP threaded code regions. So the question is: How can I more or less gracefully shut this down? How do interruption points work in an omp parallel for loop? Do I have a chance at all? My constraints here are a bit forgiving. Given the complexity of the 3rd party code I have already accepted the notion that this is gonna be a memleak monster. Luckily, there's no disk or other hardware access so I'm ready to live with a certain amount of mem not being cleaned up. Yet the backend thread communicates with the spawning class by calling a mutexed callback when it's ready so the user knows it's cancelled. This is protected by a boost mutex as well. And juding from the horrible threading "clogs" that I observed in my first trials I suppose this is gonna be a problem as well. And input is appreciated. And yes, I am aware that this is gonna be an ugly mess. Cheers, Stephan

3 3

[bimap] Mapping a set into itself
by Arindam Mukherjee 30 Sep '14

30 Sep '14

Is there a way to represent a mapping from a set to itself using boost::bimap? Regards, Arindam

1 0

choosing source ip on multihomed computer
by David Frank 30 Sep '14

30 Sep '14

Hi, I'm wondering how to make boost::asio::connect(<socket>, ...) connect from the source interface/ip address I choose on a multihomed computer. Thanks, Dave

2 1

[serialization] serialize type-erasure
by oswin krause 30 Sep '14

30 Sep '14

Hi list, i am currently trying to serialize a type-erasure, but i am not sure how this should be done exactly. Before boost 1.56 we used the polymorphic archives and a base class to implement serialization, but since they are broken in 1.56 and a few users are nagging me (plus i can't compile the library myself any more after a mandatory upgrade...), i have to get it right. my setup looks like this class ErasureBase{ .... }; template<class T> class Erasure{ public: Erasure (T& foo):m_foo(foo){} template<class Archive> void serialize(Archive & ar, const unsigned int){ ar & m_foo; } private: T& m_foo; }; class MyClass{ public: template<class T> MyClass(T& foo):m_wrapper(new Erasure<T>(foo)){} template<class Archive> void serialize(Archive & ar, const unsigned int){ ar & *m_wrapper; } private: boost::scoped_ptr<ErasureBase> m_wrapper; }; in the usage of MyClass i assume that serialisation/deserialisation is only to be called on constructed objects of type MyClass, therefore i do not need capabilities to deserialize from pointers. The question is: how do I tell the archive which Erasure<T> exist? The solutions in the tutorial pages only look at the cases where Erasure<T> is serialized directly or where it is known which classes are to be serialized (and i don't assume my users to know which Erasure<T> are instantiated...). If there exists a solution for this: is the solution also stable wrt the order of registration of classes? Best, Oswin The ques

1 1

[multi_index] Couple of questions on multi-index containers
by Aaron Levy 29 Sep '14

29 Sep '14

If I erase elements in a multi_index_container, how are iterators invalidated? 1. For other than random_access index, only iterators to the deleted element? 2. For random_access index, potentially any iterator? What are the data structures used internally for this container and its indexes. -- Aaron Levy aaron.levy(a)yandex.com

4 5

[msm] & typeerasure / any
by Raj Sambasivan 28 Sep '14

28 Sep '14

HI Boost-users/christophe, I'm trying to make use of type erasure for fsm based on suggestion from an earlier mail thread http://boost.2283326.n4.nabble.com/msm-access-to-root-fsm-td4649632.html but hitting compilation errors. Can you pls help? while the any works as expected for push back e.g for vector (bottom of email) it doesnt work for MSM.. *Doesn't Work for MSM:* BOOST_TYPE_ERASURE_MEMBER((has_process_event), process_event, 1) typedef any<has_process_event<HandledEnum(Event1 const &)>, _self&> any_fsm; struct AnyFSM { AnyFSM(any_fsm afsm_) : _afsm(afsm_) {} void process(Event1& e) { _afsm.process_event(e); } any_fsm _afsm; }; TestFSM test; test.start(); AnyFSM(test); test.process(Event1()); *error:* *test.cpp:47:43: **error: **'Event1' does not refer to a value* typedef any<has_process_event<HandledEnum(Event1 const &)>, _self&> any_fsm; *Works for vector:* BOOST_TYPE_ERASURE_MEMBER((has_push_back), push_back, 1) typedef any<has_push_back<void(int)>, _self&> any_container; void append_many(any_container container) { for(int i = 0; i < 10; ++i) container.push_back(i); } struct AnyContainer { AnyContainer(any_container cont_) : _cont(cont_) {} void append() {append_many(_cont); } any_container _cont; }; AnyContainer m(v); m.append(); for(auto i : v) { std::cout << i << std::endl; } Thanks -Raj

1 1

[Log] [ASIO with ssl] windows service crash on windows server 2008
by Pritesh Acharya 27 Sep '14

27 Sep '14

Details: OS: Windows Server 2008 SP2 Boost 1.56.0 (shared library) OpenSSL: Win32OpenSSL-1_0_1i I have a windows service application written in C++. As a minimal test case, I have created a small demo code to show the problem. The service just log some information to a file and does nothing. The same code has no problem in my Development machine(windows7) and window Server 2003. Also on Windows Server 2008, when the service is run as LocalSystem privilege, it crashes, but when run as Administrator it works. Additionally, when I remove the asio ssl header from the code, it works. or when I remove the log header, it works. The problem seems to be related to when ASIO ssl and LOG are used together. Below I've included 1. Code 2. User Mode Crash dump 3. Call Stack obtained from running Crash Dump *1. Code:* *#include <boost/log/trivial.hpp>* *#include <boost/log/sources/severity_logger.hpp>* *#include <boost/asio/ssl.hpp>* *#include <windows.h>* *#include <stdio.h>* *#define SLEEP_TIME 5000* *#define LOGFILE "C:\\AASSLInstaller_D_56\\memstatus.txt"* *SERVICE_STATUS ServiceStatus; * *SERVICE_STATUS_HANDLE hStatus; * *void ServiceMain(int argc, char** argv); * *void ControlHandler(DWORD request); * *int InitService();* *int WriteToLog(char* str)* *{* * FILE* log;* * log = fopen(LOGFILE, "a+");* * if (log == NULL)* * return -1;* * fprintf(log, "%s\n", str);* * fclose(log);* * return 0;* *}* *//void ThreadFunction()* *//{* *// WriteToLog("at ThreadFunction");* *// //std::cout << "hello World " << std::endl;* *// //getchar();* *//}* *int main() * *{ * * using namespace boost::log::trivial; //for logging* * boost::log::sources::severity_logger_mt< boost::log::trivial::severity_level > lg;* * //WriteToLog("at main");* * //return 1;* * SERVICE_TABLE_ENTRY ServiceTable[2];* * ServiceTable[0].lpServiceName = "MemoryStatus";* * ServiceTable[0].lpServiceProc = (LPSERVICE_MAIN_FUNCTION)ServiceMain;* * ServiceTable[1].lpServiceName = NULL;* * ServiceTable[1].lpServiceProc = NULL;* * // Start the control dispatcher thread for our service* * StartServiceCtrlDispatcher(ServiceTable); * * return 0;* *}* *void ServiceMain(int argc, char** argv) * *{ * * int error; * * ServiceStatus.dwServiceType = SERVICE_WIN32; * * ServiceStatus.dwCurrentState = SERVICE_START_PENDING; * * ServiceStatus.dwControlsAccepted = SERVICE_ACCEPT_STOP | SERVICE_ACCEPT_SHUTDOWN;* * ServiceStatus.dwWin32ExitCode = 0; * * ServiceStatus.dwServiceSpecificExitCode = 0; * * ServiceStatus.dwCheckPoint = 0; * * ServiceStatus.dwWaitHint = 0; * * hStatus = RegisterServiceCtrlHandler(* * "MemoryStatus", * * (LPHANDLER_FUNCTION)ControlHandler); * * if (hStatus == (SERVICE_STATUS_HANDLE)0) * * { * * // Registering Control Handler failed* * return; * * } * * // Initialize Service * * //error = InitService(); * * error = 0;* * if (error) * * {* * // Initialization failed* * ServiceStatus.dwCurrentState = SERVICE_STOPPED; * * ServiceStatus.dwWin32ExitCode = -1; * * SetServiceStatus(hStatus, &ServiceStatus); * * return; * * } * * // We report the running status to SCM. * * ServiceStatus.dwCurrentState = SERVICE_RUNNING; * * SetServiceStatus (hStatus, &ServiceStatus);* * MEMORYSTATUS memory;* * // The worker loop of a service* * //boost::thread t(&ThreadFunction);* * while (ServiceStatus.dwCurrentState == SERVICE_RUNNING)* * {* * char buffer[16];* * GlobalMemoryStatus(&memory);* * sprintf(buffer, "%d", memory.dwAvailPhys);* * int result = WriteToLog(buffer);* * if (result)* * {* * ServiceStatus.dwCurrentState = SERVICE_STOPPED; * * ServiceStatus.dwWin32ExitCode = -1; * * SetServiceStatus(hStatus, &ServiceStatus);* * return;* * }* * Sleep(SLEEP_TIME);* * }* * return; * *}* *// Service initialization* *int InitService() * *{ * * int result;* * result = WriteToLog("Monitoring started.");* * return(result); * *} * *// Control handler function* *void ControlHandler(DWORD request) * *{ * * switch(request) * * { * * case SERVICE_CONTROL_STOP: * * WriteToLog("Monitoring stopped.");* * ServiceStatus.dwWin32ExitCode = 0; * * ServiceStatus.dwCurrentState = SERVICE_STOPPED; * * SetServiceStatus (hStatus, &ServiceStatus);* * return; * * case SERVICE_CONTROL_SHUTDOWN: * * WriteToLog("Monitoring stopped.");* * ServiceStatus.dwWin32ExitCode = 0; * * ServiceStatus.dwCurrentState = SERVICE_STOPPED; * * SetServiceStatus (hStatus, &ServiceStatus);* * return; * * default:* * break;* * } * * // Report current status* * SetServiceStatus (hStatus, &ServiceStatus);* * return; * *} * *2. Dump* *Dump Summary* *------------* *Dump File: with_ssl_56.exe.4208.dmp : C:\Dumps\with_ssl_56.exe.4208.dmp* *Last Write Time: 9/26/2014 11:59:38 AM* *Process Name: with_ssl_56.exe : C:\lpa_c\wix_for_service\build_win\src\with_ssl_56\Debug\with_ssl_56.exe* *Process Architecture: x86* *Exception Code: 0xC0000005* *Exception Information: The thread tried to read from or write to a virtual address for which it does not have the appropriate access.* *Heap Information: Not Present* *System Information* *------------------* *OS Version: 6.0.6002* *CLR Version(s): * *Modules* *-------* *Module Name Module Path Module Version* *----------- ----------- --------------* *with_ssl_56.exe C:\lpa_c\wix_for_service\build_win\src\with_ssl_56\Debug\with_ssl_56.exe 0.0.0.0* *ntdll.dll C:\Windows\System32\ntdll.dll 6.0.6002.18881* *kernel32.dll C:\Windows\System32\kernel32.dll 6.0.6002.19034* *advapi32.dll C:\Windows\System32\advapi32.dll 6.0.6002.18005* *rpcrt4.dll C:\Windows\System32\rpcrt4.dll 6.0.6002.18882* *boost_log-vc100-mt-gd-1_56.dll C:\boost_1_56_0_dyn\stage\lib\boost_log-vc100-mt-gd-1_56.dll 0.0.0.0* *boost_date_time-vc100-mt-gd-1_56.dll C:\boost_1_56_0_dyn\stage\lib\boost_date_time-vc100-mt-gd-1_56.dll 0.0.0.0* *MSVCP100D.dll C:\Windows\System32\MSVCP100D.dll 10.0.30319.1* *MSVCR100D.dll C:\Windows\System32\MSVCR100D.dll 10.0.30319.1* *boost_system-vc100-mt-gd-1_56.dll C:\boost_1_56_0_dyn\stage\lib\boost_system-vc100-mt-gd-1_56.dll 0.0.0.0* *boost_filesystem-vc100-mt-gd-1_56.dll C:\boost_1_56_0_dyn\stage\lib\boost_filesystem-vc100-mt-gd-1_56.dll 0.0.0.0* *boost_chrono-vc100-mt-gd-1_56.dll C:\boost_1_56_0_dyn\stage\lib\boost_chrono-vc100-mt-gd-1_56.dll 0.0.0.0* *boost_thread-vc100-mt-gd-1_56.dll C:\boost_1_56_0_dyn\stage\lib\boost_thread-vc100-mt-gd-1_56.dll 0.0.0.0* *psapi.dll C:\Windows\System32\psapi.dll 6.0.6000.16386* *ws2_32.dll C:\Windows\System32\ws2_32.dll 6.0.6001.18000* *msvcrt.dll C:\Windows\System32\msvcrt.dll 7.0.6002.18551* *nsi.dll C:\Windows\System32\nsi.dll 6.0.6001.18000* *user32.dll C:\Windows\System32\user32.dll 6.0.6002.18005* *gdi32.dll C:\Windows\System32\gdi32.dll 6.0.6002.19171* *3. Call Stack* * 00000000() * * user32.dll!76659ce4() * * [Frames below may be incorrect and/or missing, no symbols loaded for user32.dll] * * advapi32.dll!77b84158() * * advapi32.dll!77b840f6() * * advapi32.dll!77b8430d() * * ws2_32.dll!7670ab6d() * * ws2_32.dll!7670aa95() * * ws2_32.dll!7670a3b2() * * ws2_32.dll!7670ae3f() * * ws2_32.dll!7670ad7a() * * ws2_32.dll!7670acc3() * * ws2_32.dll!7670ac2b() * * boost_log-vc100-mt-gd-1_56.dll!boost::asio::detail::winsock_init_base::startup(boost::asio::detail::winsock_init_base::data & d, unsigned char major, unsigned char minor) Line 39 + 0x2d bytes C++* * boost_log-vc100-mt-gd-1_56.dll!boost::asio::detail::winsock_init<2,0>::winsock_init<2,0>(bool allow_throw) Line 57 + 0xe bytes C++* *> boost_log-vc100-mt-gd-1_56.dll!boost::asio::detail::`dynamic initializer for 'winsock_init_instance''() Line 116 + 0x1c bytes C++* * MSVCR100D.dll!_initterm(void (void)* * pfbegin, void (void)* * pfend) Line 873 C* * boost_log-vc100-mt-gd-1_56.dll!_CRT_INIT(void * hDllHandle, unsigned long dwReason, void * lpreserved) Line 284 + 0xf bytes C* * boost_log-vc100-mt-gd-1_56.dll!__DllMainCRTStartup(void * hDllHandle, unsigned long dwReason, void * lpreserved) Line 506 + 0x11 bytes C* * boost_log-vc100-mt-gd-1_56.dll!_DllMainCRTStartup(void * hDllHandle, unsigned long dwReason, void * lpreserved) Line 476 + 0x11 bytes C* * ntdll.dll!77a112f4() * * ntdll.dll!779f906b() * * ntdll.dll!77a016d2() * * ntdll.dll!779fbf5a() *

2 1

[thread] Specific old move semantics usage
by Vicente J. Botet Escriba 27 Sep '14

27 Sep '14

Hi all, I would like to remove the specific move semantics included in Boost.Thread as it was flawed. This has been replaced by the Boost.Move one long time ago. Is there someone still using it? Could I at least define by the default (even for BOOST_THREAD_VERSION 2) BOOST_THREAD_USES_MOVE for Boost 1.57? Best, Vicente

1 0

Using boost::xpressing on Windows Store
by Paula Stanciu 27 Sep '14

27 Sep '14

Hello, I'm trying to use boost::xpressive on Windows Store and I ran into the following problem: Because _WIN32 is defined BOOST_XPRESSIVE_HAS_MS_STACK_GUARD gets defined and that causes the use of _resetstkoflw() that is not defined for windows sore ( http://msdn.microsoft.com/library/windows/apps/jj606124.aspx ) and so I cannot compile the code that uses boost::xpressive. I think BOOST_XPRESSIVE_HAS_MS_STACK_GUARD should not be defined for Windows store since it does not support _resetstkoflw() ? Any idea how I can solve this issue? Would it be the right thing to undefine this for Windows Store ? Thanks, Paula

3 2

Re: [Boost-users] Using boost::xpressive on Windows Store
by Paula Stanciu 27 Sep '14

27 Sep '14

Thanks Eric. Yes, it's if the WINAPI_FAMILY is set to WINAPI_PARTION_APP. Should I submit a patch with the changes I think should be done in this case ? It's actually just adding and extra check in xpressive_fwd.hpp before BOOST_XPRESSIVE_HAS_MS_STACK_ GUARD is defined. ( e.g. (WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_APP)) ) Best regards, Paula On Fri, Sep 26, 2014 at 6:19 PM, Eric Niebler <eniebler(a)boost.org> wrote: > On 9/26/2014 8:03 AM, Paula Stanciu wrote: > > Hello, > > > > I'm trying to use boost::xpressive on Windows Store and I ran into the > > following problem: > > > > Because _WIN32 is defined BOOST_XPRESSIVE_HAS_MS_STACK_GUARD gets > > defined and that causes the use of _resetstkoflw() that is not defined > > for windows sore ( > > http://msdn.microsoft.com/library/windows/apps/jj606124.aspx ) and so I > > cannot compile the code that uses boost::xpressive. > > > > I think BOOST_XPRESSIVE_HAS_MS_STACK_GUARD should not be defined for > > Windows store since it does not support _resetstkoflw() ? Any idea how I > > can solve this issue? Would it be the right thing to undefine this for > > Windows Store ? > > Sure. Is there some #define a program can check to see if it's being > compiled for Windows Store? > > Eric > > _______________________________________________ > Boost-users mailing list > Boost-users(a)lists.boost.org > http://lists.boost.org/mailman/listinfo.cgi/boost-users >

1 0