February 2016 - Boost-users - lists.preview.boost.org

Newbie
by Shan Wijesinghe 22 Feb '16

22 Feb '16

Hi all, I'm a new member to C++ boost and I have a fairly good knowledge in programming in C++. I'm very much excited to join this group and I need directions from the senior members on how to start myself. Please help me here. Thanks in advance. regards, DilSHAN.

6 11

Query on Boost.DLL
by Leo Carreon 21 Feb '16

21 Feb '16

Hi, I have been reading the Boost.DLL documentation. Can a single plugin support multiple API’s? If yes, when the multiple API’s are used at the same time; will there be only one instance of the plugin loaded or will there be multiple instances loaded? Regards, Leo

2 1

Boost.DLL
by Leo Carreon 21 Feb '16

21 Feb '16

Hi, Can anyone please let me know what the status of the Boost.DLL library is? Regards, Leo

2 1

[coroutine] - only std exceptions are caught from coroutines
by Tamas.Ruszkai＠leica-geosystems.com 21 Feb '16

21 Feb '16

3 4

[Interprocess] Communication between x86 and x64 issue
by Toni Neubert 19 Feb '16

19 Feb '16

I want to communicate between different platforms using boost interprocess. I am using vc12 and boost 1.58 on windows 7. My code below is a very simple example, that should work. But it doesn't for communications between different platforms... If I create it in x64 and open in win32, the process stuck at a lock at function try_based_lock in boost/int/sync/detail/common_algorithms.hpp In the other way around: win32 create, x64 open: the process crashes at name_length in segment_manager_helper.hpp while trying to find the name in index (priv_generic_find in segment_manager). #include <iostream> #include <boost/interprocess/managed_shared_memory.hpp> int main() { namespace bip = boost::interprocess; // open in WIN32, create in x64 #ifdef _WIN32 bip::managed_shared_memory msm(bip::open_only, "TestIPC"); #elsif X64 bip::shared_memory_object::remove("TestIPC"); bip::managed_shared_memory msm(bip::create_only, "TestIPC", 4096); msm.construct<uint32_t>("Data")[1](10); #endif // Get Data and print it auto data = msm.find<uint32_t>("Data"); if (data.second == 1) { std::cout << *data.first << std::endl; } std::cin.ignore(); return 0; } Does anybody have experience in this? Thank you in advance!

3 2

Re: [Boost-users] [Serialization] This compiles; but does it work?
by Bjorn Reese 19 Feb '16

19 Feb '16

On 02/19/2016 04:20 AM, Merrill Cornish wrote: > Can anyone tell me if this will work, or do I need to implement the > serialize() function? When you are doing different things for the input and output archives then it is better to split serialization into load and save: #include <boost/filesystem/path.hpp> #include <boost/serialization/split_free.hpp> namespace boost { namespace serialization { template <typename Archive> void serialize(Archive& ar, filesystem::path& p, const unsigned int version) { split_free(ar, p, version); } template<class Archive> void save(Archive & ar, const filesystem::path& p, unsigned int version) { ar << p.string(); } template<class Archive> void load(Archive & ar, filesystem::path& p, unsigned int version) { std::string data; ar >> data; p = data; } }}

2 1

[graph] Search algorithm using quality metric rather than distance
by rleigh＠codelibre.net 19 Feb '16

19 Feb '16

I have an adjacency_list graph (shown in attached dot file). Each edge has an attached "quality metric" as a bundled property. Similar to the dijkstra_shortest_paths algorithm, I'd like to determine the optimal path. However, this algorithm sums the distance along each edge, and finds the shortest path. What I'd like to do is *multiply* the quality metric for each edge in the path, and find the largest one. A perfect path will have a quality metric of 1.0, while suboptimal paths will have values tending towards zero, and the more "bad" edges the smaller the number. I'll likely then look at the number of edges and then select the shortest of the candidates. It seems like a simple algorithm that might already be included, but I just haven't found it yet not knowing its formal name. If anyone has any suggestions, I'd be very grateful. Many thanks, Roger

5 4

[Serialization] This compiles; but does it work?
by Merrill Cornish 19 Feb '16

19 Feb '16

mGraphFile is a data member of type boost:filesystem::path, which isn't (as I understand) handled natively by Boost Serialization. I could define template<class Archive> void serialize(Archive& ar, boost::filesystem::path& p, const unsigned int version); but it is needed in only one place. Below is an excerpt from an inline serialization function that I'm already using. While it has a redundant step not needed for loading and a different redundant step not needed for storing, it would seem to do what I need. (The minor inefficiency of the redundant statements is of no concern here.) ... std::string str = mGraphFile.string(); // redundant if loading ar & BOOST_SERIALIZATION_NVP(str); mGraphFile = str; // redundant if saving ... Can anyone tell me if this will work, or do I need to implement the serialize() function? Merrill Cornish

1 0

boost::msm usage: how to set properties of target states
by spiritwalker 18 Feb '16

18 Feb '16

States in my fsm has different properties. I define a event which will invoke a transition to a target state. I want to set properties of target by event data. My choices are: 1. set properties in the action or guard: but it is still in the source state. I fill not good if i set the target state properties at the time 2. set properties in the target entry: but the value of properties have been calculated in guard on receiving the event so i should recalculated again :( Any one can give me more ideas. Thanks ! -- View this message in context: http://boost.2283326.n4.nabble.com/boost-msm-usage-how-to-set-properties-of… Sent from the Boost - Users mailing list archive at Nabble.com.

2 1

SERIALIZE: Registering derived templete classes serialized via base ptr
by Unger, Jörg 17 Feb '16

17 Feb '16

I'm trying to serialize a derived class via a base ptr using boost serialize. The example to illustrate the problem (without templates) is as follows //****************** Main.cpp ********************** #include <fstream> #include <boost/archive/text_iarchive.hpp> #include <boost/archive/text_oarchive.hpp> #include "Derived.h" int main(int argc, char* argv[]) { std::string fileName = "Out.txt"; Derived myDerived; Base* myBasePtr = &myDerived; //myDerived.foo(); // serialize it as txt std::ofstream ofs(fileName.c_str(), std::ios_base::binary ); boost::archive::text_oarchive ota ( ofs, std::ios::binary ); //ota & boost::serialization::make_nvp(std::string("directly derived").c_str(), myDerived);; ota & boost::serialization::make_nvp(std::string("via base ptr").c_str(), myBasePtr);; return 0; } //*********************************************** //*************** Base.h ************************* #pragma once #include <boost/serialization/access.hpp> #include <boost/serialization/export.hpp> class Base { friend class boost::serialization::access; public: Base(){}; virtual ~Base(){}; template<class Archive> void serialize(Archive & ar, const unsigned int version) {}; }; BOOST_CLASS_EXPORT_KEY(Base) //*********************************************** //**************** Derived.h ********************** #pragma once #include <boost/serialization/access.hpp> #include <boost/serialization/export.hpp> #include "Base.h" class Derived : public Base { friend class boost::serialization::access; public: Derived(){}; void foo(); template<class Archive> void serialize(Archive & ar, const unsigned int version); }; BOOST_CLASS_EXPORT_KEY(Derived) //*********************************************** //**************** Derived.cpp ******************** #include <boost/archive/text_oarchive.hpp> #include <boost/archive/text_iarchive.hpp> #include "Derived.h" //Derived::Derived() //{} void Derived::foo() { } // serializes the class template void Derived::serialize(boost::archive::text_oarchive & ar, const unsigned int version); template void Derived::serialize(boost::archive::text_iarchive & ar, const unsigned int version); template<class Archive> void Derived::serialize(Archive & ar, const unsigned int version) { ar & BOOST_SERIALIZATION_BASE_OBJECT_NVP(Base); } BOOST_CLASS_EXPORT_IMPLEMENT(Derived) //*********************************************** When compiling the above code and linking the object files, everything works fine. When I combine everything in a library (in this example it is essentially only Derived.o) and link the main file with the library, I get an error what(): unregistered class - derived class not registered or exported The problem seems to be that the registration in the linked library is not activated. The problem is simplified here, because in our implementation, the derived class is essentially a template class that does not have any cpp-files. We have just added another otherwise empty cpp file with the registration of all possible instantiations of the class. There are several possibilities to solve the problem · The macro BOOST_CLASS_EXPORT_IMPLEMENT(Derived) could be moved to the main.cpp. This workaround is not my preferred options, since the registration has to be done in every main file we want to use the library for. · The macro BOOST_CLASS_EXPORT_IMPLEMENT(Derived) could be moved to the Base.cpp (the constructor for the base class has to be implemented in this file as well). This is another option, but this requires the header files of the derived classes to be included in Base.cpp, which is not convenient either. · Serialize one object of the derived class before serialization through the pointer, but that's only a workaround. (see the commented line in Main.cpp) · Call the foo method in Main.cpp. Surprisingly, this does not depend on whether the foo function is implemented in Derived.cpp, but also if it is called in Main.cpp. Surprisingly, if "myDerived.foo();" in Main.cpp is uncommented the program works fine, which is something I do not understand. It seems that the registration is only performed, if one of the functions in this cpp file is actually called, or am I wrong? · Implement the constructor of Derived in Derived.cpp. This approach works fine for the current example (somehow similar to the previous solution). However, in our case, Derived is a template and thus the constructor can't be implemented in the cpp file. · Another option is to always link the main file with the library and the object file Derived.o, but this is also not a convenient option. · Finally, a manual registration with register_type is possible. This could be moved to a separate subroutine that is always called in different main files. What is the recommended way of registering derived template classes that are serialized through a base ptr within a library. Best regards, Jörg F. Unger

2 1

[Timer] unresolved external symbol boost::chrono::time_point
by Aman Wardak 17 Feb '16

17 Feb '16

Env: Win7 32-bit running in VirtualBox on Macbook Pro Compiler: MS Visual Studio 14 It's my first time working with Boost. I've followed the instructions here: http://www.boost.org/doc/libs/1_60_0/more/getting_started/windows.html I've compiled Boost successfully: ============================ The Boost C++ Libraries were successfully built! The following directory should be added to compiler include paths: C:\Users\IEUser\Documents\boost_1_60_0 The following directory should be added to linker library paths: C:\Users\IEUser\Documents\boost_1_60_0\stage\lib ============================================== I've added the above paths in Visual Studio and was able to run the test in step 6 of the instructions linked above to verify my environment is set up correctly. MY PROBLEM: I'm trying to compile the very first sample code at this link: http://www.boost.org/doc/libs/1_60_0/libs/timer/doc/cpu_timers.html I get the following error: ------ Build started: Project: BoostExample, Configuration: Debug Win32 ------ Source.cpp libboost_timer-vc140-mt-gd-1_60.lib(cpu_timer.obj) : error LNK2019: unresolved external symbol "public: static class boost::chrono::time_point<class boost::chrono::steady_clock,class boost::chrono::duration<__int64,class boost::ratio<1,1000000000> > > __cdecl boost::chrono::steady_clock::now(void)" (?now@steady_clock @chrono@boost@@SA?AV?$time_point@Vsteady_clock@chrono@boost@@V?$duration@ _JV?$ratio@$00$0DLJKMKAA@@boost@@@23@@23@XZ) referenced in function __catch$??1auto_cpu_timer@timer@boost@@QAE@XZ$0 C:\Users\IEUser\Documents\Visual Studio 2015\Projects\CS2370\Debug\BoostExample.exe : fatal error LNK1120: 1 unresolved externals ========== Build: 0 succeeded, 1 failed, 0 up-to-date, 0 skipped ========== Thanks, Aman

1 0

[Test] BOOST_CHECK_EXCEPTION error message still vague
by Merrill Cornish 17 Feb '16

17 Feb '16

In BOOST_1_59_0, I complained that the error message from BOOST_CHECK_EXCEPTION was indecipherable. In BOOST_1_60_0, things improved a lot. But it still leave you hanging. I use the function fail_msg() as the predicate argument to BOOST_CHECK_EXCEPTION. If it finds a specified substring in the exception message, it returns true. Otherwise, it returns false. Below is an error message from BOOST_CHECK_EXCEPTION when the predicate returned false: exception "std::runtime_error" raised as expected: validation on the raised exception through predicate "fail_msg" The first clause says my code raised the expected exception class. That part of the message is clear and unambiguous. The second clause, however, is a non-sequitur. It seems to only acknowledge that there was a predicate named "fail_msg" without saying whether it returned true or false. Overall, the whole message doesn't say whether BOOST_CHECK_EXCEPTION passed or failed. Since a message was output, the user can assume things did not go well, but it seems like the message could be more explicit. Merrill Cornish

2 2

multiprecision max and et_on
by Daniel Brake 16 Feb '16

16 Feb '16

Hi Boost, I would like to be able to use the Eigen linear algebra library with Boost.Multiprecision with et_on. However, as noted here: https://svn.boost.org/trac/boost/ticket/11149 for max, et_on isn't compatible out of the box. Aside from writing my own little overload for max for the particular form being produced in my line of code from within Eigen, does anyone have a fix for enabling max with expression templates on? MWE: { using mpfr = boost::multiprecision::mpfr_float; // et_on by default, which is what i want using std::max; mpfr a(1), b(2), c(4); auto q = b*b+c; mpfr r = max(a,q); } Thanks a lot, Daniel

1 0

[GIL] Why doesn't pixel assignment color-convert?
by Stonewall Ballard 16 Feb '16

16 Feb '16

I've just started using boost.gil, though I've been looking at it for a while. It seems odd to me that assignment doesn't do an implicit color_convert() when the pixels are of different types. Instead, it just copies matching channels. That's the wrong thing to do when assigning an rgba pixel to an rgb pixel, since the former is straight and the latter is premultiplied. It also seems wrong that you can assign an rgba pixel to an rgb pixel, but not vice-versa. It looks like this is caused by pixels_are_compatible in gil_concept.hpp, where it fails to check that both pixels have the same number of channels. Instead, it matches channels of the output pixel against the input pixel, ignoring any extra input pixel channels. It also ignores scoped_channel_value usage. This seems like an error waiting to happen. Comments? - Stoney

1 0

ASIO: add work and remove immed., will ignore further handler_requests???
by aj neu 16 Feb '16

16 Feb '16

Hi, below is some code where boost::asio::io_service::work is added, and then immediately removed. After that, I call async_wait on a steady_timer, but it does not work: io.run() RETURNS IMMEDIATELY, and IGNORES the async_wait. Is this intended? (I was rather surprised, because I thought that the async_wait would schedule new work, but it does not work). If I cange the code-line #define WEIRD_BEHAVIOUR to #undef WEIRD_BEHAVIOUR then I have my expected behaviour. Thanks, ajneu #include <iostream> #include <chrono> #include <experimental/optional> #include <boost/asio.hpp> #include <boost/asio/steady_timer.hpp> #define WEIRD_BEHAVIOUR int main() { boost::asio::io_service io; std::experimental::optional<boost::asio::io_service::work> work; #ifdef WEIRD_BEHAVIOUR work.emplace(io); work = std::experimental::nullopt; #endif boost::asio::steady_timer timer(io, std::chrono::milliseconds{1000}); timer.async_wait([](const boost::system::error_code &ec) { std::cout << ((!ec) ? "done" : "err") << std::endl; }); io.run(); }

1 0

Writing a stratum client using boost:asio
by Jan Willem Penterman 11 Feb '16

11 Feb '16

Hi, I'm trying to write a stratum client for a cryptocurrency mining application. The mining application connects to a stratum server to receive new work to mine on and submit solutions. The stratum protocol is a simple TCP based line protocol with JSON payload. Documentation here: https://mining.bitcoin.cz/help/#!/manual/stratum-protocol I've chosen to use boost::asio, as boost is already heavily used in the fork of the project I'm building the miner for. But I'm having difficulties to understand why my current implementation doesn't work. The basic flow is as follows: 1. connect to the server. this works 2. send subscribe method. this also works 3. authorize worker(s). doesn't work. 4. stratum server should start pushing new work periodically. doesn't work. 5. submit solution when found. not implemented yet. Without authorization, the server should still start pushing work. So I've skipped step 3 for now. The source code as I have it now is available here: https://github.com/Genoil/cpp-ethereum/blob/stratum/libstratum/ The output I get from this is as follows: i 12:19:14|<unknown> Connected to stratum server coinotron.com : 3344 i 12:19:15|<unknown> {"error": null, "id": 1, "result": [["mining.notify", "91d0025c5ea54509977e3475d065169e"], "080016D0", 4]} i 12:19:15|<unknown> Subscribed i 12:19:15|<unknown> { "id": null, "method": "mining.set_difficulty", "params": [32768]} i 12:19:15|<unknown> mining.set_difficulty i 12:19:15|<unknown> {"params": ["31065", "0xc09361c52e9dddd141040cc79c52eaed382e92c1dd350324c040482a4c4ac0fa","0x5a46dd85298b0beff65ccf3006c4b5d2f7fa6ca8a5885a7f2132714fe7a48602","0x0000000200000000000000000000000000000000000000000000000000000000", true], "id": null, "method": "mining.notify"} i 12:19:15|<unknown> mining.notify >From this I can conclude that the subscription call works. Then at the end of handleResponse, I simply queue up a new async_read_until to wait for the next line of JSON to come in from the server. This also kind of works, but the idea is that after a few seconds, new work should come in as a "mining.notify". This never happens. Can anybody explain what I'm doing wrong? Regarding the mining.authorize call on line 113. I've understood from the documentation that I can send that whenever I want, so I tried to do this immediately after the subscribe method returned. But if I do this, it a) doesn't return and b) all lines after "Subscribed" don't come in any longer. This is the second problem I'm trying to figure out. I'm using this as reference, but my Python is even worse than my C++ https://github.com/slush0/stratum-mining-proxy But when I run the binaries connecting to the same host (coinotron.com:3344), I do get new work coming in... Thanks for having a look at this.

1 0

[ASIO] Newbie question - Bidirectional communication example
by Davies, John 11 Feb '16

11 Feb '16

For a simple explanation I have two processes that run on the same computer. Process 1 sends messages to Process 2 by putting the messages into an Interprocess message queue that Process 2 created. Process 2 sends messages to Process 1 by putting the messages into an Interprocess message queue that Process 1 created. The messages can be sent at any time and are not in response to each other. Very symmetrical. Now I want to put Process 1 and Process 2 on separate computers. My intent is to transfer the messages over TCP to the opposite computer without changing Process 1 or 2. So I thought I'd create two connector programs. Connector 1 creates a message queue that pretends to be the Process 2 message queue. When it gets a message, it send it over TCP the Connector 2. Connector 2 gets the message and puts it into the real Process 2 message queue. And the reverse if Process 2 has an outgoing message. I just can't find a good example to steal from. Everything I can find are more like Process 1 makes a request and Process 2 responds. So I think I have a fundamental misunderstanding of how ASIO works. ________________________________ The information contained in this message may be confidential and legally protected under applicable law. The message is intended solely for the addressee(s). If you are not the intended recipient, you are hereby notified that any use, forwarding, dissemination, or reproduction of this message is strictly prohibited and may be unlawful. If you are not the intended recipient, please contact the sender by return e-mail and destroy all copies of the original message.

2 1

[Test] Funny Business with vector::at() out-of-range
by Merrill Cornish 11 Feb '16

11 Feb '16

I posted a question a couple of days ago: "Throws 'Escaping' both BOOST_CHECK_THROW and try{}". I found that BOOST_CHECK_EXCEPTION which is working for me elsewhere fails the same way. So, an out-of-range std::vector::at() seems to be the common element. In the definition of at(), the out-of-range response is handled by __throw_out_of_range_fmt(). When I looked up that one, I found void __throw_out_of_range_fmt(const char*, ...) __attribute__((__noreturn__)) __attribute__((__format__(__gnu_printf__, 1, 2))); in functexcept.h. To my admittedly less-than-expert eye, there doesn't seem to be an actual throw() here. On the other hand, the entire file are entries like this for various exceptions. There are a lot of Google hits for __throw_out_of_range_fmt, but they are for unresolved references in links, apparently caused by mixed version of libraries and object. Everything I've got is compiled linked for MinGW 64-bit, GCC 5.3.0, 64-bit --std=C++14, BOOST 1_60_1 64-bit so I get none of those errors. The problem I'm seeing with my test code is that wrapping a try{} around BOOST_CHECK_THROW shows that [1] the body of the try doesn't complete; but at the same time [2] not even a catch(...) can catch the presumed throw--as if a throw isn't actually being done. So, where did the throw go? Merrill Cornish

2 1

[TEST] Throws "Escaping" both BOOST_CHECK_THROW and try{}
by Merrill Cornish 11 Feb '16

11 Feb '16

I'm running Windows 10, 64-bit; MinGW 64-bit; GCC 5.3 64-bit; and BOOST 1_60_1 Here is the relevant test code: [319] BOOST_CHECK_THROW(throw std::out_of_range("BOOST_CK_THROW test"), std::out_of_range); [320] try { [321] std::cerr << "arcsSize()" << std::endl; [322] empty1.arcsSize(Operand::DEFAULT_ARC_SET+1); [323] std::cerr << "arcsSize() did not throw" << std::endl; [324] }//try [325] catch (const std::out_of_range& ex) [326] { std::cerr << "out_of_range caught=|" << ex.what() << "|" << std::endl; } [327] catch (...) { std::cerr << "something else caught" << std::endl; } [328] [329] BOOST_CHECK_THROW(empty1.arcsSize(Operand::DEFAULT_ARC_SET+1), std::out_of_range); The original code was just the last line, line [329]. [319]-[327] was added in an attempt to characterize the problem. The error [329] got by itself was: terminate called after throwing an instance of 'std::out_of_range' what(): vector::_M_range_check: __n (which is 1) >= this->size() (which is 1) This application has requested the Runtime to terminate it in an unusual way. Please contact the application's support team for more information. SIGABRT intercepted; converted to exit(EXIT_FAILURE). arcSize() is implemented with std::vector::at() and the index is out-of-range, so it's supposed to throw std::out_of_range. That's what was detected, so why did it terminate the program? (The last line comes from my SIGABRT handler. Without it, the test executable ends up locked and can't be deleted.) Adding in all of the test lines shown above gives; [319] Operand_Test.cpp(319): info: check 'exception "std::out_of_range" raised as expected' has passed [321] arcsSize() terminate called after throwing an instance of 'std::out_of_range' what(): vector::_M_range_check: __n (which is 1) >= this->size() (which is 1) This application has requested the Runtime to terminate it in an unusual way. Please contact the application's support team for more information. SIGABRT intercepted; converted to exit(EXIT_FAILURE). Notice the explicit throw in line [319] is detected and handled properly. However, the arcSize() call from inside the try block throws, but is not caught by even catch(...) {}. Can anyone help me see what's going on here? Merrill Cornish

2 6

[Container] Regarding Trac #11957, static_vector::max_size() is higher than the capacity
by Gagan Kumar 10 Feb '16

10 Feb '16

I tried a simple fix for https://svn.boost.org/trac/boost/ticket/11957 here https://github.com/boostorg/container/pull/37/files . Can someone please review it. Regards,Gagan Kumar

1 0

Boost Variant Type Interfaces
by Ali Nakipoglu 09 Feb '16

09 Feb '16

Hi, In our project we are using boost, boost.variant very extensively. We was using methods provided by the library to interface with Variant types. Mainly : get<T>( variant ) for accessing stored values and variant.which() == Index for checking currently stored type. Overtime we realized we can improve this as : variant.getT() for accessing stored values much more nicer to write and variant.isT() or variant.which() == WHICH_T for checking currently stored type which does not depend on order of template paramters in the Variant definition. Also in some cases we wanted to be able to add type specific methods for example in our case we are using subscript operator to access recursively stored Variants variant[ "a" ].isT() or variant[ 0 ].asT() We generalized this solution for our project and published at https://github.com/FictionIO/WrappedVariant I would like to hear your comments. Do you think its a correct approach or useful? Any suggestions are welcome! Kind Regards, Ali

1 0

[Multiprecision] Runtime assertion failure with clang++3.6.0 but not g++4.9.2
by mbw_bst＠mailbox.org 09 Feb '16

09 Feb '16

Dear all, this is my first post on this mailing list, so hopefully this will all be by the book. Executing the following minimal working example, which is supposed to calculate "3! == 6" via (1) a template function (2) a class template and save the result in a boost::multiprecision::mpz_int, results in unexpected behavior for clang++3.6.0, with the version of my boost library being 1.59.0. The program displays the correct result using g++4.9.2 however: #include <iostream> #include <boost/multiprecision/gmp.hpp> namespace bm = boost::multiprecision; template <std::size_t N> bm::mpz_int factorial() { return N * factorial<N-1>(); } template<> bm::mpz_int factorial<0>() { return 1; } template <std::size_t N> struct factorial2 { static const bm::mpz_int value; }; template<std::size_t N> const bm::mpz_int factorial2<N>::value = N * factorial2<N-1>::value; template<> struct factorial2<0> { static const bm::mpz_int value; }; const bm::mpz_int factorial2<0>::value = 1; int main () { std::cout << factorial<3>() << '\n'; std::cout << factorial2<3>::value << '\n'; return 0; } The output for clang, having been compiled via clang++ -std=c++14 -stdlib=libc++ -DNDEBUG -o main main.cpp -lgmp will result in: 6 0 However, omitting the NDEBUG flag will trigger an assertion failure: main: /usr/include/boost/multiprecision/gmp.hpp:1296: const mpz_t &boost::multiprecision::backends::gmp_int::data() const: Assertion `m_data[0]._mp_d' failed. Aborted (core dumped) Compiling the same program with g++4.9.2 (with or without "-DNDEBUG") yields the expected result: 6 6 What did I miss? Can this behavior indeed be traced down to Boost.Multiprecision, or does it have to do with the initialization order of static const class members? Or something else? Additionally, both variants work correctly with clang when using a built-in type like 'int'. Any help in understanding the problem on hand would be greatly appreciated! Best Regards, mbw

3 5

Re: [Boost-users] [Multiprecision] Runtime assertion failure with clang++3.6.0 but not g++4.9.2
by mbw_bst＠mailbox.org 09 Feb '16

09 Feb '16

Hello again and thanks for your replies! Taking the following two statements into account: > For type int, the values are computed recursively at compile time, for a > user-defined-type the values are computed at runtime and you hit > initialization-order issues. > Seems to be that Multiprecision relies on static data that haven't > (necessarily) been initialized before your own static members. Works > fine > if the members are made non-static: Does that mean that, since a) boost::multiprecision::number<gmp_int> is not constexpr enabled, and b) static initialization can't happen at compile-time for user-defined types, that there is no way for me to initialize a variable of this type at compile-time? Does b) follow logically from a) ? I (believe to have) confirmed runtime-initialization by running valgrind, which showed the presence of dynamic memory allocations (and also jumps depending on uninitialized variables, which probably corresponds to the assertion failure). I am mentioning this because initially my little test program looked somewhat like this: #include <iostream> #include <vector> #include <boost/multiprecision/gmp.hpp> namespace bm = boost::multiprecision; template<std::size_t N> struct factorial { static const bm::mpz_int value; static void add_values(std::vector<bm::mpz_int>& vec) { factorial<N-1>::add_values(vec); vec.push_back(value); } }; template<std::size_t N> const bm::mpz_int factorial<N>::value = N * factorial<N-1>::value; template<> struct factorial<std::size_t{0}> { static const bm::mpz_int value; static void add_values(std::vector<bm::mpz_int>& vec) { vec.push_back(value); } }; const bm::mpz_int factorial<0>::value = 1; int main () { std::vector<bm::mpz_int> fac_table{}; factorial<10'000>::add_values(fac_table); std::cout << "Enter a number between 0 and 10'000 to calculate factorial:"; std::size_t number; std::cin >> number; std::cout << '\n' << number << "! == " << fac_table[number] << '\n'; return 0; } This still includes the "unsafe" static initialization of a user-defined type. However it compiles and runs via g++ -ansi -std=c++14 -O3 -Wall -Wextra -pedantic -static -save-temps -ftemplate-depth-50000 -DNDEBUG -o main main.cpp -lgmp -pthread This was what made me believe in the compile-time computation of the factorials in the first place, since using this version, valgrind does not show any dynamic memory allocation at all. On the other hand, I don't understand what's going on here really, especially since std::vector<> is supposed to allocate memory at run-time. Now I'm not sure whether the last example has anything to do with Boost.Multiprecision anymore, so while it would really make my day if you could help me better understand this, I will understand if this is not the right place to ask these questions. Thanks again for your replies so far, they have been of great help! Cheers, Max

2 1

[gsoc16] ATTN: Project ideas deadline 19th Feb
by Niall Douglas 07 Feb '16

07 Feb '16

Tomorrow 8th February the application period for Google Summer of Code 2016 begins, and yet we have just *one* idea on the ideas page: https://svn.boost.org/trac/boost/wiki/SoC2016 We need *plenty* more if we are to be approved again this year as a Google Summer of Code org, the deadline for which is the 19th February. So if you have some project ideas and think you could mentor a student this summer, *please* consider adding a project idea to the ideas list or stepping forward with a list of Boost libraries or topics you'd be willing to mentor. Our thanks in advance. --- Boost C++ Libraries Google Summer of Code 2016 admin https://svn.boost.org/trac/boost/wiki/SoC2016

1 0

download from sourceforge?
by Dan Bloomquist 07 Feb '16

07 Feb '16

I followed the link form boost.org and then through sourceforge. But sourceforge directed me to: <http://downloadbud.com/?adprovider=sourceforge_fwint&source=fw36_300x250-do…> The link is an executable! I've never heard of 'downloadbud'. Not a big deal for now, I'm running with 1.58. But I'd like to get the latest of spirit x3 before I ask questions on the spirit list that may not need asking. Is there another source for boost? Or, is this executable actually a safe extractor? Sourceforge handing over an exe when it should be be a package seems just wrong... Thanks, Dan.

2 5