- Boost-users - lists.boost.org

14 Mar '12

Hi all, I have had a problem with Boost 1.42 on Solaris 10 an SunStudio 12 and found out, that there was a relation to a bug #3886. I do the required patch what I found there and the compilation worked afterwards. My question: Because we use this boost version for different platforms, how can I notice, if this patch is only for this platform or also for others. At least we use our configuration management for this issue and I have to now, if I can checkin this patch for all or only for some or at least only for this special platform. Can somebody give me an answer to that? regards Arno

1 0

[MPL] Fold a vector of pairs
by Peter Nyssen 14 Mar '12

14 Mar '12

Nevermind. Removing the typename before mpl::fold fixes it. Kind regards, Peter //------------------------------------- #include <boost/mpl/if.hpp> #include <boost/mpl/int.hpp> #include <boost/mpl/fold.hpp> #include <boost/mpl/pair.hpp> #include <boost/mpl/push_back.hpp> #include <boost/mpl/vector.hpp> namespace mpl = boost::mpl; struct tag1 {}; struct tag2 {}; template <typename T> struct is_tag1 : boost::false_type {}; template <> struct is_tag1<tag1> : boost::true_type {}; int main() { typedef mpl::vector< mpl::pair<tag1, mpl::int_<1> >, mpl::pair<tag2, mpl::int_<2> >, mpl::pair<tag1, mpl::int_<3> > > initial_vector; typedef mpl::fold< initial_vector, mpl::pair<mpl::vector<>, mpl::vector<> >, mpl::if_< is_tag1<mpl::first<mpl::_2> >, mpl::push_back< mpl::first<mpl::_1>, mpl::second<mpl::_2> >, mpl::push_back< mpl::second<mpl::_1>, mpl::second<mpl::_2> > > >::type sorted_pair; return 0; }

1 0

[property_tree] if "ostream << string_path" missing ?
by Alain O Miniussi 14 Mar '12

14 Mar '12

Hi, I cannot find a direct way to output a string_path on an ostream. Is the feature simply missing or considered useless, and in this case, what is the rationale ? Thanks Alain

2 1

build error using sharable_lock with -02 optimization
by Magallo 14 Mar '12

14 Mar '12

I'm trying to use the Boost.Interprocess library, mutexes, scoped_lock, sharable_lock etc... If I use a simple test program that uses mutex like this, I have no problem: void test1() { printf("test1 - timed_lock \n"); printf("trying to get exlusive lock within a timeout of 5 seconds\n"); boost::interprocess::named_mutex mutex(boost::interprocess::open_or_create, "MyExlusiveTimedMutex"); boost::posix_time::ptime abs_time = boost::posix_time::second_clock::universal_time() + boost::posix_time::seconds(5); if( mutex.timed_lock(abs_time) ) { printf("mutex acquired! do some processing...\n"); //...do some processing... printf("unlocking mutex, quit.\n"); mutex.unlock(); } else { printf("mutex not acquired, quit.\n"); } printf("end of test1\n"); } This test is ok, it builds with no problem. Then I try to use the scoped_lock functionalities: void test2() { //open or create the named mutex printf("creating mutex...\n"); boost::interprocess::named_mutex n_mutex(boost::interprocess::open_or_create, "ipc_pri_queue_mutex"); { printf("acquiring mutex scoped_lock...\n"); boost::interprocess::scoped_lock<boost::interprocess::named_mutex> sc_lock(n_mutex); printf("mutex scoped_lock acquired...\n"); //wait some time to simulate execution printf("execute program...\n"); sleep(10); printf("program excuted...\n"); } printf("out of scope, sharable_lock released\n"); } Also this test is ok, it builds with no problem. Finally, I try to use the sharable_lock and I get the error: void test3() { //open or create the named upgradable mutex printf("creating mutex...\n"); boost::interprocess::named_upgradable_mutex n_u_mutex(boost::interprocess::open_or_create, "ipc_pri_queue_upgradable_mutex"); { printf("acquiring mutex sharable_lock...\n"); boost::interprocess::sharable_lock<boost::interprocess::named_upgradable_mutex> sh_lock(n_u_mutex); printf("mutex sharable_lock acquired...\n"); //wait some time to simulate execution printf("execute program...\n"); sleep(10); printf("program excuted...\n"); } printf("out of scope, sharable_lock released\n"); } If I build this with -O2 optimization flag, I get this error: g++ -O2 -I /usr/local/boost_1_49_0 -c main.cpp -o main.o /usr/local/boost_1_49_0/boost/interprocess/detail/atomic.hpp: In member function 'void boost::interprocess::ipcdetail::managed_open_or_create_impl<DeviceAbstraction, MemAlignment, FileBased, StoreDevice>::priv_open_or_create(boost::interprocess::ipcdetail::create_enum_t, const typename boost::interprocess::managed_open_or_create_impl_device_id_t<DeviceAbstraction>::type&, size_t, boost::interprocess::mode_t, const void*, const boost::interprocess::permissions&, ConstructFunc) [with ConstructFunc = boost::interprocess::ipcdetail::named_creation_functor<boost::interprocess::interprocess_upgradable_mutex, boost::interprocess::ipcdetail::named_creation_functor_no_arg>, DeviceAbstraction = boost::interprocess::shared_memory_object, unsigned int MemAlignment = 0u, bool FileBased = true, bool StoreDevice = true]': /usr/local/boost_1_49_0/boost/interprocess/detail/atomic.hpp:316: internal compiler error: in rewrite_stmt, at tree-into-ssa.c:1380 Please submit a full bug report, with preprocessed source if appropriate. See <URL:http://forum.java.sun.com/forum.jspa?forumID=905> for instructions. *** Error code 1 make: Fatal error: Command failed for target `main.o' If I build this without the -O2 flag, the build is ok. Why this? Please help. I did this test on: OS: Solaris 10 5/09 s10s_u7wos_08 SPARC Compiler: sparc-sun-solaris2.10-g++ (GCC) 4.2.1 (20090222) (gccfss)

1 0

[MPL] Fold a vector of pairs
by Peter Nyssen 14 Mar '12

14 Mar '12

Hi all, I'm trying to understand how to use mpl::fold to transform a vector like this: vector< pair<tag1, A> pair<tag2, B> pair<tag1, C> > into the following: pair< vector<A,C> vector<B> > Following test program fails to compile and I don't understand why. If you could explain me what I'm missing here, you will gain my eternal respect. Kind regards, Peter //-------------------------------------------- #include <boost/mpl/if.hpp> #include <boost/mpl/int.hpp> #include <boost/mpl/fold.hpp> #include <boost/mpl/pair.hpp> #include <boost/mpl/push_back.hpp> #include <boost/mpl/vector.hpp> namespace mpl = boost::mpl; struct tag1 {}; struct tag2 {}; template <typename T> struct is_tag1 : boost::false_type {}; template <> struct is_tag1<tag1> : boost::true_type {}; int main() { typedef mpl::vector< mpl::pair<tag1, mpl::int_<1> >, mpl::pair<tag2, mpl::int_<2> >, mpl::pair<tag1, mpl::int_<3> > > initial_vector; typedef typename mpl::fold< initial_vector, mpl::pair<mpl::vector<>, mpl::vector<> >, mpl::if_< is_tag1<mpl::first<mpl::_2>::type >, mpl::push_back< mpl::first<mpl::_1>::type, mpl::second<mpl::_2>::type >, mpl::push_back< mpl::second<mpl::_1>::type, mpl::second<mpl::_2>::type > > >::type sorted_pair; return 0; } //-------------------------------------------- *$* g++ -Wall -I ../boost/ -o test test.cpp In file included from test.cpp:5:0: ../boost/boost/mpl/pair.hpp: In instantiation of 'boost::mpl::first<mpl_::arg<2> >': test.cpp:32:40: instantiated from here ../boost/boost/mpl/pair.hpp:43:31: fout: no type named 'first' in 'struct mpl_::arg<2>' test.cpp: In functie 'int main()': test.cpp:32:47: fout: template argument 1 is invalid In file included from test.cpp:5:0: ../boost/boost/mpl/pair.hpp: At global scope: ../boost/boost/mpl/pair.hpp: In instantiation of 'boost::mpl::first<mpl_::arg<1> >': test.cpp:34:36: instantiated from here ../boost/boost/mpl/pair.hpp:43:31: fout: no type named 'first' in 'struct mpl_::arg<1>' ../boost/boost/mpl/pair.hpp: In instantiation of 'boost::mpl::second<mpl_::arg<2> >': test.cpp:35:37: instantiated from here ../boost/boost/mpl/pair.hpp:56:32: fout: no type named 'second' in 'struct mpl_::arg<2>' test.cpp: In functie 'int main()': test.cpp:36:13: fout: template argument 1 is invalid test.cpp:36:13: fout: template argument 2 is invalid In file included from test.cpp:5:0: ../boost/boost/mpl/pair.hpp: At global scope: ../boost/boost/mpl/pair.hpp: In instantiation of 'boost::mpl::second<mpl_::arg<1> >': test.cpp:38:37: instantiated from here ../boost/boost/mpl/pair.hpp:56:32: fout: no type named 'second' in 'struct mpl_::arg<1>' test.cpp: In functie 'int main()': test.cpp:40:13: fout: template argument 1 is invalid test.cpp:40:13: fout: template argument 2 is invalid test.cpp:41:9: fout: template argument 1 is invalid test.cpp:41:9: fout: template argument 2 is invalid test.cpp:41:9: fout: template argument 3 is invalid test.cpp:42:5: fout: template argument 3 is invalid test.cpp:42:24: fout: invalid type in declaration before ';' token **

2 1

Re: [Boost-users] [Graph] On Dijkstra SSSP performance
by Leo Hidd 13 Mar '12

13 Mar '12

sry, I could not make a reply in the post through http://thread.gmane.org/gmane.comp.lib.boost.user/73084/focus=73086 If i use the "Follow up" action and post it I have this message: "You seem to be top-posting. Don't do that." That's why I'm doing this now. Hope it works. Anyways... thx Jeremiah for the inputs. I tested what you said for a mesh of 3 million vertices (then, 9 million edges), and here is a summary of the different configs: 1- graph type: out-edge as listS; predecessors: vector<vertex_descriptor> p(num_vertices(g)); distances: vector<double> d(num_vertices(g)); NDEBUG defined (release configuration); no optimizations: - dijkstra_shortest_paths elapsed time: 11970ms 2- graph type: out-edge as listS; predecessors: vertex_descriptor* p = new vertex_descriptor[num_vertices(g)]; distances: double* d = new double[num_vertices(g)]; NDEBUG defined (release configuration); no optimizations: - dijkstra_shortest_paths elapsed time: 9446ms 3- graph type: out-edge as vecS; predecessors: vertex_descriptor* p = new vertex_descriptor[num_vertices(g)]; distances: double* d = new double[num_vertices(g)]; NDEBUG defined (release configuration); no optimizations: - dijkstra_shortest_paths elapsed time: 8436ms 4- graph type: out-edge as listS; predecessors: vertex_descriptor* p = new vertex_descriptor[num_vertices(g)]; distances: double* d = new double[num_vertices(g)]; NDEBUG defined (release configuration); with optimizations (option /Ox) - dijkstra_shortest_paths elapsed time: 5741ms 5- my implementation without optimizations takes 3289ms and with optimizations takes 2860ms. Using predecessors and distances storage as simple arrays instead of std::vector have a real gain (although optimization does even more). Is it possible to use arrays (maybe array of arrays) instead of vecS for OutEdgeList and VertexList? Or put arrays inside some kind of class with functions allowing compatibility/interfacing with the Dijkstra algorithm (like insert, sort, remove, or whatever it needs)? Since the Dijkstra algorithm does more read access to the edge list than the distance or path arrays, it could really accelerate things. thx again, leo

2 10

error compiling under Solaris10 in release mode
by Magallo 13 Mar '12

13 Mar '12

Hi all, I am new to boost libraries and I find them very powerful. I'm using the Boost.Interprocess library to make some interprocess communication and synchronization. I'm using mutexes and I was able to let my program build and work as expected. This is good when I build it in Debug mode, when I switch to release mode, i.e. using the -O2 or -O3 flag to optimize performances, the compiler returns this error: "boost/interprocess/detail/atomic.hpp:316: internal compiler error: in rewrite_stmt, at tree-into-ssa.c:1380" Please submit a full bug report, with preprocessed source if appropriate. When I use a debug compilation, without -O2, it works fine. Why that? I read somewhere that I could try to define BOOST_SP_USE_PTHREADS but I tryed and nothing changes. Is it something related to the operating system? I see that the error brings me to inline boost::uint32_t atomic_cas32 (volatile boost::uint32_t *mem, boost::uint32_t with, boost::uint32_t cmp) { return __sync_val_compare_and_swap(const_cast<boost::uint32_t *>(mem), cmp, with); } in the atomic.hpp file. The compiler is: g++ --version sparc-sun-solaris2.10-g++ (GCC) 4.2.1 (20090222) (gccfss) Copyright (C) 2007 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. What could I do? Please help. Cheers.

1 0

[Graph] How to do multi source shortest path
by Leo Hidd 13 Mar '12

13 Mar '12

Hi all, I have a graph and multiple sources (please read this thread: http://thread.gmane.org/gmane.comp.lib.boost.user/73084/focus=73086 first paragraph, for the characteristics of the graph, if it matters). So, instead of running the single source shortest path (SSSP) algorithm (e.g., dijkstra_shortest_paths()) multiple times, what I do is that I specify several sources instead of only one (so all the sources are initialized with distance 0 and put into the queue) and than I run Dijkstra as one usually does for SSSP. Doing so, the multi source runs almost as fast as a single source instead of running NbOfSouces*ElapsedTimeForSingleSource. Then, I'd like to know if it's possible to use boost algorithms for multi-source shortest path, like specifying several sources or anything alike. And, in the case it is possible, it would be possible to have an array with the index of the source the closest to a given node? All pairs is not a solution because the number of sources is way less than the number of nodes in the graph and because all sources use lots of memory for storing the distances (for a 4 million nodes graph I'd need a 4*10^12 elements table, don't?) thx for your time, leo

4 14

Re: [Boost-users] Boost interprocess containers - performance issue
by Naveen Santhapuri 12 Mar '12

12 Mar '12

> Date: Sun, 11 Mar 2012 12:21:36 +0100 > From: Ion Gazta?aga <igaztanaga(a)gmail.com> > To: Boost User List <boost-users(a)lists.boost.org> > Subject: Re: [Boost-users] Boost interprocess containers - performance issue > Message-ID: <4F5C8AC0.9020508(a)gmail.com> > Content-Type: text/plain; charset=ISO-8859-1; format=flowed >El 11/03/2012 1:00, Naveen Santhapuri escribi?: >> Hello all, >> >> I am experimenting with Boost interprocess containers and like the >> ease of using them. I want to use these in place of the "linux shared >> memory+shared memory allocator + std containers". However, I am seeing >> some performance hit when using boost. I measured the times for vector >> access from shared memory. I define access as indexing into the vector >> (v[i]) (all times in milli seconds): >> >> Num elements Local heap | Boost ipc | Shmat() >> 100,000 0.23 0.39 0.27 >> 1,000,000 2.3 3.85 2.7 > What does Shmat mean? Are you putting a std::vector directly in a shared > memory segment? > Obviously if you put boost containers in shared memory you don't store > raw pointers (as every process can map the memory in a different > address, making raw pointers unusable). When using offset_ptr we have a > performance hit as we need to calculate the real address computing a > pointer arithmetic (this + stored offset). Apart from that, as > offset_ptr relies on undefined behaviour, the offset calculation must be > marked as non-inlineable in most compilers, increasing the performance > hit. I think it can be improved using some kind of compiler-specific > annotations or assembler, but I have no skills to do that nowadays. > > Best, > > Ion Thank you for the reply Ion. Re: shmat() - Yes, I am creating a std::vector with shared allocator in shared memory. My use case: One executable (lets call it shvec) writes std containers to shared memory. Another instance of shvec reads those containers from shared memory. I want to do this to preload a large amount of data into memory, before my service starts, so that the start time of the service can be cut down. I want to use boost but I am afraid of the runtime hit. Even with the offset pointer, do the numbers above make sense (as much as 70% extra time)? My alternative strategy without boost ipc: I will be running on 64bit linux and hoping to map the shared memory segment to the same location in all instances using shmat(). To achieve that, the first instance gets the shared memory segment address, which will be used by the other instances that read from the shared memory. If I can map them to the same address, can I use raw pointers without any problem? Naveen

2 1

Boost timer initialization hangs in win_therad.ipp
by chaitanya sonavale 12 Mar '12

12 Mar '12

Hi, I have a C++ DLL which has exported functions which is used in a C# application via DllImport. The C++ DLL is linked to various .lib files which use boost functionality (thread, asio etc) internally and are compiled with BOOST_ALL_DYN_LINK option. When I run the application it just hangs with following stack trace. I get the same results if I convert this to a C++\CLI DLL and use it directly from the C# application. Stack Trace: ------------------------------------------------ OneComShim.dll!boost::asio::detail::win_thread::start_thread(boost::asio::detail ::win_thread::func_base * arg=0x035574f0, unsigned int stack_size=0x00010000) Line 98 + 0xe bytes C++ OneComShim.dll!boost::asio::detail::win_thread::win_thread<boost::asio::detail:: win_iocp_io_service::timer_thread_function> (boost::asio::detail::win_iocp_io_service::timer_thread_function f={...}, unsigned int stack_size=0x00010000) Line 72 C++ OneComShim.dll!boost::asio::detail::win_iocp_io_service::do_add_timer_queue(boos t::asio::detail::timer_queue_base & queue={...}) Line 471 + 0x31 bytes C++ OneComShim.dll!boost::asio::detail::win_iocp_io_service::add_timer_queue<boost:: asio::time_traits<boost::posix_time::ptime> > (boost::asio::detail::timer_queue<boost::asio::time_traits<boost::posix_time::pt ime> > & queue={...}) Line 78 C++ OneComShim.dll!boost::asio::detail::deadline_timer_service<boost::asio::time_tra its<boost::posix_time::ptime> >::deadline_timer_service<boost::asio::time_traits<boost::posix_time::ptime> > (boost::asio::io_service & io_service={...}) Line 68 C++ OneComShim.dll!boost::asio::deadline_timer_service<boost::posix_time::ptime,boos t::asio::time_traits<boost::posix_time::ptime> >::deadline_timer_service<boost::posix_time::ptime,boost::asio::time_traits<boos t::posix_time::ptime> >(boost::asio::io_service & io_service={...}) Line 73 C++ OneComShim.dll!boost::asio::detail::service_registry::create<boost::asio::deadli ne_timer_service<boost::posix_time::ptime,boost::asio::time_traits<boost::posix_ time::ptime> > >(boost::asio::io_service & owner={...}) Line 81 + 0x2f bytes C++ OneComShim.dll!boost::asio::detail::service_registry::do_use_service(const boost::asio::io_service::service::key & key={...}, boost::asio::io_service::service * (boost::asio::io_service &)* factory=0x100c770d) Line 123 + 0xc bytes C++ OneComShim.dll!boost::asio::detail::service_registry::use_service<boost::asio::d eadline_timer_service<boost::posix_time::ptime,boost::asio::time_traits<boost::p osix_time::ptime> > >() Line 49 C++ OneComShim.dll!boost::asio::use_service<boost::asio::deadline_timer_service<boos t::posix_time::ptime,boost::asio::time_traits<boost::posix_time::ptime> > > (boost::asio::io_service & ios={...}) Line 34 C++ OneComShim.dll!boost::asio::basic_io_object<boost::asio::deadline_timer_service< boost::posix_time::ptime,boost::asio::time_traits<boost::posix_time::ptime> > >::basic_io_object<boost::asio::deadline_timer_service<boost::posix_time::ptime, boost::asio::time_traits<boost::posix_time::ptime> > >(boost::asio::io_service & io_service={...}) Line 91 + 0x4b bytes C++ OneComShim.dll!boost::asio::basic_deadline_timer<boost::posix_time::ptime,boost: :asio::time_traits<boost::posix_time::ptime>,boost::asio::deadline_timer_service <boost::posix_time::ptime,boost::asio::time_traits<boost::posix_time::ptime> > >::basic_deadline_timer<boost::posix_time::ptime,boost::asio::time_traits<boost: :posix_time::ptime>,boost::asio::deadline_timer_service<boost::posix_time::ptime ,boost::asio::time_traits<boost::posix_time::ptime> > >(boost::asio::io_service & io_service={...}) Line 147 C++ OneComShim.dll!CORE::SERVER_COMMUNICATOR::CurlHandler::CurlHandler(CORE::SERVER_ COMMUNICATOR::ServerAgent * serverAgent=0x1029c950) Line 37 + 0x118 bytes C++ OneComShim.dll!CORE::SERVER_COMMUNICATOR::ServerAgent::ServerAgent() Line 22 C++ OneComShim.dll!`dynamic initializer for 'CORE::SERVER_COMMUNICATOR::ServerAgent::agent''() Line 17 + 0x28 bytes C++ TestWin.exe!TestWin.Program.Main(string[] args = {string[0x00000000]}) Line 16 + ------------------------------------------------ As it can be seen, it stops in start_thread inside win_thread.ipp at following lines on WaitForSingleObject if (entry_event) { ::WaitForSingleObject(entry_event, INFINITE); ::CloseHandle(entry_event); } I am just creating a timer object in the code and do not understand why it leads to this situation. Also everything works fine when I use a LIB instead of DLL and all code is in C++. Environment details are as follows OS: Win XP IDE: Visual Studio 2010 Boost version: boost_1_48_0 Regards, Chaitanya

1 0