September 2006 - Boost-users

boost+stlport+Xcode
by Don Agro 22 Sep '06

22 Sep '06

Has anyone had any luck linking boost into an XCode application that is also using STLport ? I am using STLport 5.0.0 and boost 1.33.1 Boost links in fine unless I also link in STLport then I have one undefined symbol... boost::signals::detail::signal_base::signal_base (boost::function2<bool, boost::signals::detail::stored_group, boost::signals::detail::stored_group, stlp_std::allocator<boost::function_base> > const&, boost::any const&) bjam build (with STLport at "/STLport-5.0.0" bjam -sSTLPORT_PATH="/" -sSTLPORT_VERSION="5.0.0" -sTOOLS="darwin" Any ideas ? Thanks in advance.

2 2

[boost::filesystem] path namecheck : want to use "native" on Windows ?
by Washburn 22 Sep '06

22 Sep '06

I'm using Boost on Windows and the filesystem::path behaviour seems incorrect with the default name check, so I want to use the "native" namecheck instead. First of all, is this normal ? (since Boost is compiled for windows platformn should'nt the right name check be set by default ?) I initialize it manually using : filesystem::path::default_name_check( filesystem::native ) But since it can only be done once, this throws exceptions occasionnaly (not very important, but "ugly") Is there a better way to set the name check at the start of the application ? Washburn

3 2

boost/pool/detail/mutex.hpp and threads on HP-UX
by mojmir.svoboda＠logicacmg.com 22 Sep '06

22 Sep '06

hello, i'm trying to build opendiameter (uses boost) with gcc 3.4.3 on a hp box: HP-UX bruno B.11.23 U 9000/800 1532750548 unlimited-user license and i've encountered following error when compiling libdiamparser: boost_1_33_1/boost/pool/detail/mutex.hpp:41:4: #error Unable to determine platform mutex type; define BOOST_NO_MT to assume single-threaded well, according to <sys/unistd.h> _POSIX_THREADS is still undefined on HP-UX: /************************************************************************ * The following defines are specified in the standard, but are not yet * implemented: * _POSIX_THREADS can't be defined until all features are implemented ... i've found similar problem in python and the solution uses check for _SC_THREADS instead of _POSIX_THREADS. yet better solution may use BOOST_HAS_PTHREADS: boost_1_33_1/boost/pool/detail $ diff -Naur mutex.hpp.orig mutex.hpp --- mutex.hpp.orig 2006-09-19 14:20:53.000000000 +0200 +++ mutex.hpp 2006-09-19 16:11:50.000000000 +0200 @@ -31,7 +31,7 @@ #define BOOST_MUTEX_HELPER BOOST_MUTEX_HELPER_WIN32 #else #include <unistd.h> - #ifdef _POSIX_THREADS + #ifdef BOOST_HAS_PTHREADS #define BOOST_MUTEX_HELPER BOOST_MUTEX_HELPER_PTHREAD #endif #endif @@ -45,7 +45,7 @@ # ifdef BOOST_WINDOWS # include <windows.h> # endif -# ifdef _POSIX_THREADS +# ifdef BOOST_HAS_PTHREADS # include <pthread.h> # endif #endif @@ -83,7 +83,7 @@ #endif // defined(BOOST_WINDOWS) -#ifdef _POSIX_THREADS +#ifdef BOOST_HAS_PTHREADS class pthread_mutex { @@ -107,7 +107,7 @@ { pthread_mutex_unlock(&mtx); } }; -#endif // defined(_POSIX_THREADS) +#endif // defined(BOOST_HAS_PTHREADS) #endif // !defined(BOOST_NO_MT) because the file <boost/config/platform/hpux.hpp> defines BOOST_HAS_PTHREADS for us on the hpux+gcc. is it allright? thanks fya, mojmir This e-mail and any attachment is for authorised use by the intended recipient(s) only. It may contain proprietary material, confidential information and/or be subject to legal privilege. It should not be copied, disclosed to, retained or used by, any other party. If you are not an intended recipient then please promptly delete this e-mail and any attachment and all copies and inform the sender. Thank you.

2 1

boost::interprocess::managed_external_buffer - link errors
by alexmark 21 Sep '06

21 Sep '06

When I include boost/interprocess/managed_external_buffer.hpp in more than one file within a solution (MSVC 7.1) I get linker errors LNK2005 and LNK4006 because of boost::interprocess::fill_system_message(...) function redefinition. The same is applicable to the case when this header file is included in stdafx.h only. fill_system_message(...) is a standalone function, not a member of a class. And the body of it is included in header file. May it be the point the problem originates from? Maybe the better way is to move "fill_system_message(...)" to .cpp source file? Have anybody faced this problem? Thanks, Alex.

2 1

How to do code factoring when using Boost.Program_options?
by Peng Yu 21 Sep '06

21 Sep '06

Hi, Suppose that I have program option set U. I have many programs which uses a subset of U as their options. Some of them share the same subset. For now, I code all the options in the main function like the one in the tutorial http://www.boost.org/doc/html/program_options/tutorial.html However, there is a problem when I have a lot programs with similar options. I would like to factor out the common code. But I don't have an idea how to do it in a good way. If you happen to have some experience, please let me know and share your experience with me. Thanks, Peng

2 2

[boost::signal][boost::bind] signal::disconnect problem with MSVC 2005
by david callu 21 Sep '06

21 Sep '06

Hi, See this code ... #include<iostream> #include<boost/bind.hpp> #include<boost/signal.hpp> struct CSignal { typedef boost::signal< void (void) > UpdateSignal; UpdateSignal sig; }; struct CSlot {: void update(int i) { std::cout << "Call void update( " << i << " )" << std::endl; } void connect(CSignal& cs) { cs.sig.connect(boost::bind(&CSlot::update, this, 5)); } void disconnect(CSignal& cs) { cs.sig.disconnect(boost::bind(&CSlot::update, this, 5)); } }; int main() { CSignal csignal; CSlot cslot; cslot.connect(csignal); csignal.sig(); cslot.disconnect(csignal); csignal.sig(); return (0); } When i compile with GCC 4.1.1, this compile fine. But when i compile with MSVC 2005 it report an error: .\main.cpp(30) : error C2664: 'boost::signal1<R,T1,Combiner,Group,GroupCompare,SlotFunction>::disconnect' : cannot convert parameter 1 from 'boost::_bi::bind_t<R,F,L>' to 'const int &' with [ R=void, T1=int, Combiner=boost::last_value<void>, Group=int, GroupCompare=std::less<int>, SlotFunction=boost::function<void (int)> ] and [ R=void, F=boost::_mfi::mf1<void,CSlot,int>, L=boost::_bi::list2<boost::_bi::value<CSlot *>,boost::arg<1>> ] Reason: cannot convert from 'boost::_bi::bind_t<R,F,L>' to 'const int' with [ R=void, F=boost::_mfi::mf1<void,CSlot,int>, L=boost::_bi::list2<boost::_bi::value<CSlot *>,boost::arg<1>> ] No user-defined-conversion operator available that can perform this conversion, or the operator cannot be called have you any idee to solve this problem? thank david callu

3 4

Boost.signals with -pthread
by Pavel Syomin 21 Sep '06

21 Sep '06

Hi, all! I am novice in this list, so don't beat freely :) I confronted with a strange thing using boost.signals on SLES 9 SP3. I don't shure Boost.signals bug, becase on FC4 and RHEL AS 3 all fine. Test case: --------------------- CUT HERE ------------------ #include <boost/signal.hpp> #include <iostream> struct MySlot { void operator() () { std::cout << "YES!" << std::endl; } }; MySlot myslot; boost::signal<void ()> mysig; int main() { mysig.connect(myslot); return 0; } --------------------- CUT HERE ------------------ Compile and run: c++ test.cc -o test -lboost_signals ./test Programm will succesfully exit. Compile and run: c++ test.cc -o test -lboost_signals -pthread ./test Programm will not exit never. Boost version: 1.31.0 (from SuSE rpm) GCC version: c++ (GCC) 3.3.3 (SuSE Linux) I run this test on other SuSE boxes with same results. Any ideas? P.S. I can't use GCC without '-pthread' because my program use threads...

2 5

[Boost Conference] Save the Date
by David Abrahams 21 Sep '06

21 Sep '06

================================ Save the Date: May 14-18, 2007 ================================ We're pleased to issue this pre-announcement of the first annual Boost Conference, BoostCon 2007. The conference will take place in Aspen, Colorado, May 14-18. This promises to be the main face-to-face event for all things Boost, from from using Boost libraries to writing them, from evangelizing Boost to deployment within your organization, from infrastructure and process to vision and mission, and from TR1 to TR2. Given the range and interests of the participants, this event promises to be intense and in-depth. Due to the breadth of our community, the conference will be organized into two primary tracks: one directed mostly at hard-core Boost developers, and the other focused on the practical interests of Boost end-users. In the spirit of Boost, however, the conference will be planned to foster collaboration and interaction both within *and* across these tracks, with an emphasis on hands-on, participatory sessions. More details, including a call for session proposals, will be forthcoming in the next few weeks. Enrollment for this event will be limited, so we're issuing this notice to the Boost community well in advance. In the meantime, we have set up a page on the Boost Wiki where we're doing an informal, preliminary survey of interest (http://www.crystalclearsoftware.com/cgi-bin/boost_wiki/wiki.pl?BoostCon) If you want to be a part of the inaugural BoostCon and are likely to attend, we'd appreciate it if you'd add yourself (or an alias) to the list there, to help us with event planning. Thanks, The BoostCon 2007 Organizers ------------------------------------------------------------------ David Abrahams Beman Dawes Jeff Garland Kevlin Henney Scott Meyers Eric Niebler Sean Parent Rene Rivera Jeremy Siek Matthias Troyer ------------------------------------------------------------------

2 2

Boost/ptr_vector and BCB6
by Ivo Trowitzsch 21 Sep '06

21 Sep '06

Hello everyone, I am new to Boost. I am using Borland C++ Builder 6 (bcc32 v5.6). Now I wanted to use the ptr_container/ptr_vector: So I included <boost/ptr_container/ptr_vector.hpp>. that's where the trouble starts, without actually using a ptr_vector. Compiling leaves me with errors in reversible_ptr_container.hpp: "reversible_ptr_container.hpp(487): E2404 Dependant type qualifier 'reversible_ptr_container<Config,CloneAllocator>' has no member type named 'allow_null'." To be honest: I have no clue, what to do to to get rid of that error. I guess ptr_vector should work on the bcc32, at least I was not able to find anything related to this error searching this mailing-list's archive and elsewhere with google. I imagine I might have set some compiler's settings wrong or something, but I don't know. If anyone could help me out on this, I'd be very thankful! Ivo.

2 1

Probably bug in condition::timed_wait()
by Pavel Syomin 21 Sep '06

21 Sep '06

Hi, I'm porting custom library from UNIX to Windows and found that some piece of code work correctly on UNIX and some time segfaulting on Windows. I think, that problem is on boost::condition::timed_wait() function. To be sure I write small test: #include <stdexcept> #include <iostream> #include <boost/thread.hpp> #include <windows.h> using namespace boost; using namespace std; class AsyncQueue { public: AsyncQueue() : m_counter(0) {} void push() { mutex::scoped_lock lock(m_mutex); m_counter++; m_condition.notify_one(); } void timed_pop(const xtime &time) { mutex::scoped_lock lock(m_mutex); if(m_counter) { m_counter--; return; } if(!m_condition.timed_wait(lock, time)) return; if(!m_counter) throw out_of_range("oops..."); m_counter--; } private: mutex m_mutex; condition m_condition; int m_counter; }; static AsyncQueue queue; static void start() { for(;;) { queue.push(); Sleep(100); } } int main(int argc, char **argv) { thread thread(start); xtime time; for(;;) { xtime_get(&time, TIME_UTC); time.nsec += 10000; queue.timed_pop(time); } return 0; } AsyncQueue is a simple asynchronous queue container with two operations - push() and timed_pop(). Because this is test, AsyncQueue don't really collect something, but it's has variable (m_counter field) that count current number on elements in container. When push() called m_counter incremented and when timed_pop() "virtually" remove item from queue m_counter decremented. In main() I create additional thread, which push data to queue. Data popped from queue in main thread by calling timed_pop(). But sometimes timed_pop throw out_of_range exception. If you look in code you can see that it's mean that condition::timed_wait() return true, but queue is empty! I look to boost-users list and found e-mail's about similar problems from Eric Colleu, but decision wasn't found. Two words about my environment. I use Visual Studio 2005, Windows XP SP2 on VMWare 5.5.1. Boost version is 1.33.1. Thanks for constructive answers.

3 5