September 2014 - Boost-users

boost.python built with VC2010
by Mohammed El-Afifi 25 Sep '14

25 Sep '14

Hello, I have successfully built boost.python with visual c++ 2010(dynamic library, 64-bit). I have python 2.7.8 64-bit installed on windows 7 64-bit. I've read here https://docs.python.org/2/distutils/apiref.html#module-distutils.msvccompil… that python extensions should be built with the same compiler that was used to build python itself. My question is, since python 2.7.8 was built with visual c++ 2008 64-bit compiler, can I have the following mix a) my python extension built with VC2008 64-bit(to match python itself) b) 64-bit boost.python dll built with VC2010 64-bit c) python 64-bit, built itself with VC2008 64-bit? I've noticed that boost.python dll depends on python dll and that's why I'm quite worried about the mix. Appreciating your insights.

1 0

[boost/config] BOOST_LIKELY argument requirements
by Stephan Bergmann 25 Sep '14

25 Sep '14

Looking at the definition of BOOST_LIKELY as #define BOOST_LIKELY(x) __builtin_expect(x, 1) in boost/config/compiler/{clang,intel,gcc}.hpp, and given that at lest the GCC and Clang __builtin_expect has a first parameter of type long, I wonder (1) whether it would be better to define it as #define BOOST_LIKELY(x) __builtin_expect(bool(x), 1) so that > $ cat test.cc > #include "boost/config.hpp" > #include "boost/logic/tribool.hpp" > int f(boost::tribool b1, boost::tribool b2) { > if (b1) return 1; > else if (BOOST_LIKELY(b2)) return 2; > else return 3; > } > > $ g++ -c test.cc > In file included from .../boost_1_56_0/boost/config.hpp:39:0, > from test.cc:1: > test.cc: In function ‘int f(boost::logic::tribool, boost::logic::tribool)’: > .../boost_1_56_0/boost/config/compiler/gcc.hpp:90:46: error: cannot convert ‘boost::logic::tribool’ to ‘long int’ for argument ‘1’ to ‘long int __builtin_expect(long int, long int)’ > #define BOOST_LIKELY(x) __builtin_expect(x, 1) > ^ > test.cc:5:14: note: in expansion of macro ‘BOOST_LIKELY’ > else if (BOOST_LIKELY(b2)) return 2; > ^ would not fail (effectively changing the requirement for the argument to BOOST_LIKELY from "[the] expression should result in a boolean value" to "it should be contextually convertible to bool"). (2) if not that, whether to at least wrap the expansion of x in parentheses, #define BOOST_LIKELY(x) __builtin_expect((x), 1) Stephan

1 0

[assign][ptr_container] Exception safety of ptr_list_of, ptr_push_back, ptr_map_insert, etc.
by Aaron Levy 25 Sep '14

25 Sep '14

I am trying to understand how ptr_list_of or ptr_push_back provide more exception safety. I can write something like: boost::ptr_vector<Foo> vec; push_back(vec)(new Foo)(new Foo); Is the order of evaluation of the two "new Foo" expressions indeterminate here? If not, then if the second new throws, the destructor of vec should be invoked. The Foo object allocated by the first new and now part of vec should also get released as part of that. -- Aaron Levy aaron.levy(a)yandex.com

3 3

[Log] reversed file rotation order...
by Ovanes Markarian 24 Sep '14

24 Sep '14

Hi *, We use file rotation with boost log, since I work more and with Python I use Python's logging as well. One interesting feature I observed was that Python's logger does reversed file rotation. Is that sort of feature implemented with boost::log library? In boost the rotation is: whatever_0.log < oldest file whatever_1.log whatever_2.log < newest file In Python the rotation order is: whatever_0.log < newest file whatever_1.log whatever_2.log < oldest file I think this is a very cool how Python does it, since I am able to endlessly "tail" in the shell the log file: tail -f whatever_0.log will not stop until I cancel the tail operation. With boot it is like, logger writes a new file and tail hangs with the file which is no longer written. This is a bit unusable and confusing. Would be great if that'd be possible in boost. Thanks, Ovanes

1 0

Problems Linking with Boost Serialization
by Isaac Lascasas 24 Sep '14

24 Sep '14

I have found this on boost 1.56 with msvc12: Error 119 error LNK2005: "bool __cdecl boost::archive::detail::is_whitespace<char>(char)" (??$is_whitespace@D @detail@archive@boost@@YA_ND@Z) already defined in libboost_wserialization-vc120-mt-1_56.lib(basic_text_wiprimitive.obj) C:\supernova\sb6\src\supernova_clr\libboost_serialization-vc120-mt-1_56.lib(basic_text_iprimitive.obj) Error 120 error LNK2005: "bool __cdecl boost::archive::detail::is_whitespace<wchar_t>(wchar_t)" (??$is_whitespace@ _W@detail@archive@boost@@YA_N_W@Z) already defined in libboost_wserialization-vc120-mt-1_56.lib(basic_text_wiprimitive.obj) C:\supernova\sb6\src\supernova_clr\libboost_serialization-vc120-mt-1_56.lib(basic_text_iprimitive.obj) This code used to link fine with previous boost versions. Now it looks like you can't use both archive and warchive at the same time. I'm using xml warchives for serializing in my code. Regards. Isaac Lascasas.

2 2

Boost::Locale jam file seems to be outdated for ICU ver. 53.1
by Caroline Beltran 24 Sep '14

24 Sep '14

I'm using Boost::Locale (and ICU) for the first time but I strongly believe that the jam file is referencing old ICU dll filenames. Current ICU dlls: icuucd53.dll icudt53.dll icuind53.dll icuin53.dll Relevant snippet from jam file (Boost 1.56): searched-lib icuuc : : <toolset>msvc <variant>debug <name>icuucd <search>$(ICU_PATH)/lib <link>shared <runtime-link>shared ; To me it appears that the snippet above is referring to icuucd when it should be referring to icuucd53. I've been trying really hard to get Boost::Locale going on my system. I updated all references to dlls and this is the result of the config test. As you can see, I am not linking with ICU (even though ICU was installed correctly and ICU test results were successful). I think that the jam file may need further changes. Please advise. D:\boost_1_56_0\stage\lib>test_config - Backends: std winapi - Without iconv - Environment LANG= LC_ALL= LC_CTYPE= TZ= - C locale: English_United States.1252 - C++ locale: English_United States.1252 - Testing locales availability on the operation system: locale C C++ en_US.UTF-8 No No en_US.ISO8859-1 No No English_United States.1252 Yes Yes he_IL.UTF-8 No No he_IL.ISO8859-8 No No Hebrew_Israel.1255 Yes Yes ru_RU.UTF-8 No No Russian_Russia.1251 Yes Yes tr_TR.UTF-8 No No Turkish_Turkey.1254 Yes Yes ja_JP.UTF-8 No No ja_JP.SJIS No No Japanese_Japan.932 Yes Yes - Testing timezone and time Local Time :%c=09/23/14 15:18:07; %Z=Central Daylight Time (Mexico); %z=Central Daylight Time (Mexico) Universal Time:%c=09/23/14 20:18:07; %Z=Central Standard Time (Mexico); %z=Central Standard Time (Mexico) - Boost.Locale's locale: en_US.UTF-8

2 1

Adding more than 20 rows in transition table for Boost::MSM
by Yechiel Levi 23 Sep '14

23 Sep '14

Hi, I tried defining: #define BOOST_MPL_CFG_NO_PREPROCESSED_HEADERS #define BOOST_MPL_LIMIT_VECTOR_SIZE 30 // or whatever you need #define BOOST_MPL_LIMIT_MAP_SIZE 30 // or whatever you need /// Boost-MSM Includes // back-end #include "boost/mpl/vector/vector30.hpp" #include <boost/msm/back/state_machine.hpp> //front-end #include <boost/msm/front/state_machine_def.hpp> #include <boost/msm/front/functor_row.hpp> #include <boost/any.hpp> // for And_ operator #include <boost/msm/front/euml/operator.hpp> And i still get the following errors: builder1: In file included from /usr/local/include/boost/mpl/erase.hpp:19, builder1: from /usr/local/include/boost/fusion/mpl/erase.hpp:10, builder1: from /usr/local/include/boost/fusion/mpl.hpp:20, builder1: from /usr/local/include/boost/fusion/include/mpl.hpp:11, builder1: from /usr/local/include/boost/msm/back/state_machine.hpp:30, builder1: from File_Meta_FSM.h:18, builder1: from Open_Files_Mng.h:5, builder1: from Open_Files_Mng.cpp:1: builder1: /usr/local/include/boost/mpl/aux_/erase_impl.hpp:56: error: ‘_’ was not declared in this scope builder1: /usr/local/include/boost/mpl/aux_/erase_impl.hpp:56: error: ‘_’ was not declared in this scope builder1: /usr/local/include/boost/mpl/aux_/erase_impl.hpp:56: error: template argument 1 is invalid builder1: /usr/local/include/boost/mpl/aux_/erase_impl.hpp:56: error: template argument 2 is invalid builder1: /usr/local/include/boost/mpl/aux_/erase_impl.hpp:57: error: template argument 3 is invalid builder1: /usr/local/include/boost/mpl/aux_/erase_impl.hpp:62: error: ‘_’ was not declared in this scope builder1: /usr/local/include/boost/mpl/aux_/erase_impl.hpp:62: error: ‘_’ was not declared in this scope builder1: /usr/local/include/boost/mpl/aux_/erase_impl.hpp:62: error: template argument 1 is invalid builder1: /usr/local/include/boost/mpl/aux_/erase_impl.hpp:62: error: template argument 2 is invalid builder1: /usr/local/include/boost/mpl/aux_/erase_impl.hpp:63: error: template argument 3 is invalid and more of the same errors.. -- Kind regards, Yechiel Levi

5 7

Graph concepts for Dijkstra shortest-path searches
by Clayton Davis 22 Sep '14

22 Sep '14

Hi, I am somewhat uncertain on which concepts are required for the Dijkstra searches in the BGL and parallel BGL. For the sequential BGL, documentation at http://www.boost.org/doc/libs/1_56_0/libs/graph/doc/dijkstra_shortest_paths… would seem to indicate that a VertexListGraph is required, even when the no_init version is used. Looking at the code, I think that is accurate - use of the index map at https://github.com/boostorg/graph/blob/master/include/boost/graph/dijkstra_… seems to require vertex list traversal - but I would also think that defeats the purpose of the no_init option. Conversely, for the parallel BGL documentation at http://www.boost.org/doc/libs/1_56_0/libs/graph_parallel/doc/html/dijkstra_… says only that DistributedGraph is required, which is a refinement of Graph but not VertexListGraph or IncidenceGraph. However, just a little bit of searching seems to show that the parallel Dijkstra search requires both, along with perhaps PropertyGraph (see for instance https://github.com/boostorg/graph_parallel/blob/master/include/boost/graph/… and https://github.com/boostorg/graph_parallel/blob/master/include/boost/graph/…) I won't pretend to understand the code; but I would expect these functions to require something stronger than the documentation indicates. Could someone please clarify what the requirements are for these functions? Any help is greatly appreciated. Thanks, Clayton

1 0

BOOT_PP_REPEAT simple nesting problem
by Colin Fowler 22 Sep '14

22 Sep '14

Hi all, I'm trying to use BOOST_PP to generate some code. I can generate the sequence int x_0 to int x_4 using the code below. #include <boost/preprocessor/repetition/repeat.hpp> #define DECL1(z, n, text) text ## _ ## n = n; #define decl1(text, count) BOOST_PP_REPEAT(count, DECL1, text) decl1 (int x, 5) What I'd like to do is write a decl2 that can be used something like this: decl2 (int x, 5, 2) to generate int x_0_0 to int x_4_1 I've tried nesting BOOST_PP_REPEAT but so far I've been unable to produce the desired behaviour unless I hardcode one of the iteration counts. I'd like to also eventually write a decl3(int x, 5, 3, 4) when I understand how a decl2 would work. Any help much appreciated! Regards, Colin

4 6

[assign][cref_list_of] Creating anonymous list of pairs
by Aaron Levy 22 Sep '14

22 Sep '14

Is there a way to generate a list of value pairs using ref_list_of or cref_list_of. Maybe such a list could be assigned to a map. Something like: ref_list_of<size, std::pair<std::string, std::string>>("foo", "bar")("foo1", "bar1")...; or std::map<string, string> names = ref_list_of<size, std::pair<std::string, std::string>>("foo", "bar")("foo1", "bar1")...; -- Aaron Levy aaron.levy(a)yandex.com

3 2

[ptr_container][ptr_map]
by Aaron Levy 22 Sep '14

22 Sep '14

Consider the following code: class Base { public: virtual ~Base() = 0 {}; virtual double get() = 0; }; class D1 : public Base { public: D1(double n) : val_(n) {} double get() const { return val_; } private: double val_; }; class D2 : public Base { D2(double n) : val_(n) {} double get() const { return val_; } private: double val_; }; bool operator < (const Base& left, const Base& right) { return (left.get() < right.get()); } I can create a ptr_set of Base like this: boost::ptr_set<Base> set; But I can't create a ptr_map with Base as key, like this: boost::ptr_map<Base, Base> map; The problem is that Base is abstract and so the key cannot be instantiated by ptr_map. What is the cause for this distinction between ptr_set and ptr_map. -- Aaron Levy aaron.levy(a)yandex.com

2 1

Boost.Test multiple linkage question
by Rob Harris 22 Sep '14

22 Sep '14

All, I've been ripping out my hair all day, and I can't seem to determine a straight answer from the Googles. Can I link _multiple_ boost unit test dynamic libraries into a single binary and have all of them detected? For instance... src lib liba foo.cpp liba_unittests.cpp libb bar.cpp libb_unittests.cpp libd baz.cpp libd_unittests.cpp lib liba.so liba_ut.so libb.so libb_ut.so libd.so libd_ut.so bin ut unittest_driver When I build this infrastructure out, everything compiles. I've checked the compilation command and it looks like everything is on the command line, but there's explicit indication that unittest_driver actually read or linked any of the lib*_ut.so files. If I make individual unit test drivers in, say, liba, the driver works fine. If I link everything into a single unittest_driver binary, I keep getting the dreaded empty test tree. I have -DBOOST_ALL_DYN_LINK set in my CXXFLAGS so that shouldn't be the problem. And I only have BOOST_TEST_MAIN set in the driver; still no joy. Any help (or a definitive, "no you idiot, you can't do that!") would be greatly appreciated. Thanks. -RMH

2 1

Utah C++ User Group started
by legalize+jeeves＠mail.xmission.com 22 Sep '14

22 Sep '14

[Please do not mail me a copy of your followup] I'm booting up a C++ User Group in Utah. We're bootstrapping out first meeting to decide a regular meeting time and location. Join the Utah C++ Programmers MeetUp group to hear about meetings: <http://www.meetup.com/Utah-Cpp-Programmers/> Follow related evens on the Utah C++ Users Group blog: <http://utahcpp.wordpress.com> -- "The Direct3D Graphics Pipeline" free book <http://tinyurl.com/d3d-pipeline> The Computer Graphics Museum <http://computergraphicsmuseum.org> The Terminals Wiki <http://terminals.classiccmp.org> Legalize Adulthood! (my blog) <http://legalizeadulthood.wordpress.com>

1 0

[lockfree] stack needs undocumented trivial assignment
by Norbert Wenzel 21 Sep '14

21 Sep '14

I'm trying to replace some computation intensive loops in legacy code with parallel loops and thus tried to store the results in a lockfree stack. I've chosen stack over queue because of the requirements stated in the documentation, where stack only needs a copy ctor and queue additionally needs trivial assignment and trivial dtor. My result struct I want to store in the stack is *not* trivially assignable. When trying to compile the code containing my lockfree::stack the static assertion in line 75 of stack.hpp because it requires trivial assignment and trivial dtor. My question would be if this is simply an error in the documentation ie. if the requirements section should be copied over from lockfree::queue or if these static assertions should only be checked in some special case, eg. where the stack is of fixed size? thanks, Norbert

1 0

[context] How to set stack pointer in a platform-independent way?
by TONGARI J 21 Sep '14

21 Sep '14

Hi, The problem had been discussed here: https://groups.google.com/forum/#!topic/boost-devel-archive/cl8B1oUPrVM But it's still not clear how to set the stack pointer portably, even after examining the code of Boost.Coroutine. For now, Boost.Context only supports downward stack, and Boost.Coroutine, which happened to be written by the same author, just assumes that the stack always grows downward, which is not really a portable implementation, if future Boost.Context supports upward stack, Boost.Coroutine has to be changed as well. Even than that, what does "beginning of the stack" really means? Suppose we have a stack of size 3, starting at [s]: [s-1][s][s+1][s+2][s+3] Why the beginning of a downward stack is [s+3], not [s+2]? If it's a upward stack, is beginning [s] or [s-1]? Thanks.

5 12

from: giridhar
by giridhar 20 Sep '14

20 Sep '14

Salutations boost http://ariasking.com/morning.php/ttqhx3144vcfnpry/ giridhar

1 0

[ASIO] [SSL] [SOCKS4] Layers
by Maciej Miszczak 19 Sep '14

19 Sep '14

Hello, I have been successfully using ASIO for a long time, but recently I try to use SSL over SOCKS 4 and I'm struggling with short read while asynchronous SSL handshaking. I have come across this: http://stackoverflow.com/questions/9943541/boost-asio-how-to-read-write-on-…] but it's only partially helpful. I haven't find any good explanation, what the layers really are (ex. I can't assign next_layer and lowest_layer to any of OSI layers). Could someone explain the meaning of next_layer and lowest_layer and how to use SSL stream with SOCKS 4 proxy. Best regards, Maciej Miszczak

2 2

Re: [Boost-users] Boost XML Serialization
by Manasi Deshpande 19 Sep '14

19 Sep '14

Hello All, Currently I am working on DB Driven C++ API, where it also involves XML processing functionalities. That's where I came across Boost Serialization. We have a very peculiar story here, and it goes like this: We have 2 identical APIs. One in C# and another in C++. Both are DB driven similar APIs. There are instances, where any XML data serialized on C++ platform needs to be de serialized on C# and vice versa. In C# life is straight forward; one needs no bothering about serialization and de serialization. I think in boost you can de serialize the XML document only if it is built using Boost Serialization. I think it is the case in C# also. Is there any way out, where Boost Serialization can be customized? Please forgive me if I am not clear in stating my problem statement! -------------------------------------------------------------------------------------------------- This email and any attachments are confidential and are for the use of the addressee only. If you are not the addressee, you must not use or disclose the contents to any other person. Please immediately notify the sender and delete the email. Statements and opinions expressed here may not represent those of the company. Email correspondence is monitored by the company. This information may be subject to export control regulation. You are obliged to comply with such regulations. Renishaw plc (company number 1106260) and Wotton Travel Limited (company number 01973158) are companies registered in England and Wales with a registered office at New Mills, Wotton-under-Edge, Gloucestershire, GL12 8JR, United Kingdom, Telephone +44 1453 524524. --------------------------------------------------------------------------------------------------

2 1

[logging+unittest]
by Witold E Wolski 19 Sep '14

19 Sep '14

Hi, I need to integrate some loggin into my unit to some extend bein integration tests (hardware). How best to call this function which sets up boost logging void init() { boost::log::add_file_log("sample.log"); boost::log::core::get()->set_filter ( boost::log::trivial::severity >= boost::log::trivial::info ); } if I have a test Runner looking like: #define BOOST_TEST_MODULE MasterTestSuite #include <boost/test/unit_test.hpp> Is there a way not have to add the suites manually Thanks. -- Witold Eryk Wolski

1 1

[thread] memory synchronization multiple mutexes
by firespot 19 Sep '14

19 Sep '14

Hi, How does memory synchronization and visibility behave when multiple mutexes are involved? I have two groups of threads (boost::thread_group) and two mutexes (boost::mutex m1, boost::mutex m2) + associated condition variables, with each thread group being responsible to write to some variables (here Ptr, run) and consistently using "its" mutex for locking. However there is a piece of code that looks - much simplified - like this: void f() { { boost::unique_lock<boost::mutex> Lock(m1); Ptr = someAddress; // set the Ptr from 0 to some memory address } { boost::unique_lock<boost::mutex> Lock(m2); run = 1; // was 0, inform that execution can proceed condVar2.notify_all(); } ... // wait until Ptr has been reset to 0 { boost::unique_lock<boost::mutex> Lock(m1); while (Ptr != 0) condVar1.wait(Lock); } } and another function: void g() { { boost::unique_lock<boost::mutex> Lock(m2); while (run == 0) condVar2.wait(Lock); } // run the execution, e.g. std::copy(Ptr, Ptr + 100, Target); // finally all finished, signal back that Ptr is not needed any more { boost::unique_lock<boost::mutex> Lock(m1); Ptr = 0; condVar1.notify_all(); } } f() is only invoked from a thread of the first group, and g() only from a thread of the second group. Whenever Ptr might get changes the first mutex (m1) and its associated condition-variable is used, and whenver run might get changed the second mutex and its associated conditional variable is used. The control flow guarantees that run is not set to 1 prior to Ptr being set, and thus in g() the "running" itself (here the std::copy) cannot commence earlier. However, what puzzles me now: I used a different mutex for setting Ptr and setting/checking run. Does this have any impact whatsoever on memory visibility across the threads in the two thread-groups? Specifically, is it always guaranteed that when g() continues its execution (because run == 1) that it must see the correct value of Ptr (a non-NULL value)? Or is it possible that it still sees 0 and I'd have to make a separate locking involving m1 to guarantee a proper read for Ptr as anything written under a lock of m1 is also only guaranteed to be later properly read if the locking also used m1? thanks!

2 1

[mpl] simple mpl::transform with placeholder?
by Andrew King 19 Sep '14

19 Sep '14

Hi, Lets say I have an mpl vector defined as : template <typename Traits> struct A { }; template <typename Traits> struct B { }; typedef mpl::vector<A<mpl::_>, B<mpl::_> > MyVec; struct MyTraits {}; How do I 'apply' or transform MyVec with MyTraits? ie. I want the mpl::vector ... boost::mpl::vector<A<MyTraits>, B<MyTraits>>; I realise I can do mpl::apply<A<mpl::_>, MyTraits> to get A<MyTraits> but I can't work out how to do the equivalent to the entire vector. mpl::transform seems like what I should be using. Transform seems to do apply internally, but I can't work out the syntax to do something like: using substituted = typename boost::mpl::transform<MyVec, boost::mpl::apply<boost::mpl<_>, MyTraits> >::type; (This isn't right!) It may be I am confused because inside the transform, boost::mpl::_1 is the item in the vector, not the 'inner' placeholder. Sorry if this is a basic question ... Regards, Andrew

1 0

[Thread] Read-write locks
by lfrfly＠icqmail.com 18 Sep '14

18 Sep '14

I'm trying to create a standard read/write lock using Boost Threads, and I'm having a bit of difficulty mapping the Boost concepts onto the usual ones. A read/write lock should have several states: Locked for reading (multiple threads may lock at once), locked for writing (only one thread may lock, and no threads may be locked for reading at the same time), and unlocked. Ideally it would be nice if an RWL which is locked for reading may temporarily upgrade to locked for writing, then return to the locked for reading state. Boost supports several relevant notions of locking: shared locking, upgrade locking, and exclusive locking. This to me is a confusing distinction. In particular, a shared_mutex seems appropriate as the basis for a RWL, but is lock_upgrade() the equivalent of a write lock and lock_shared the equivalent of a read lock? This is unclear. If "upgrade locking" is not the same as "exclusive locking", then is an upgrade lock actually exclusive? Could someone please clarify this.

4 3

Numeric Cast from integer to floating point
by dariomt＠gmail.com 17 Sep '14

17 Sep '14

Hi, Quoting the docs for boost::numeric_cast: "There are several situations where conversions are unsafe: [...] Conversions from floating point types to integral types." What about conversions from integral types to floating point types? E.g. from 64bit int to double. The following example shows what I mean: #include <iostream> #include <cmath> #include <boost/numeric/conversion/cast.hpp> int main() { const uint64_t i = 123445678911188878; std::cout << "i=" << i << std::endl; const double d = i; std::cout << "d=" << std::fixed << d << std::endl; std::cout << "next=" << std::fixed << std::nextafter(d, std::numeric_limits<double>::max() ) << std::endl; std::cout << "prev=" << std::fixed << std::nextafter(d, -std::numeric_limits<double>::max()) << std::endl; // I'd expect the following cast to fail const double dd = boost::numeric_cast<double>(i); std::cout <<"dd=" << std::fixed << dd << std::endl; return 0; } prints i=123445678911188878 d=123445678911188880.000000 next=123445678911188896.000000 prev=123445678911188864.000000 dd=123445678911188880.000000 because that integer cannot be represented in double precision Is there something in Boost to help here? Thanks in advance

5 4

value_type in Typed In Place Factory
by Adam Badura 17 Sep '14

17 Sep '14

According to header file boost/utility/typed_in_place_factory.hpp (as linked from documentation: http://www.boost.org/doc/libs/1_55_0/boost/utility/typed_in_place_factory.h…) typed_in_place_factory has member type value_type which is the constructed type. Yet documentation ( http://www.boost.org/doc/libs/1_55_0/libs/utility/in_place_factories.html) fails to mention it. Why? Is it not part of the interface (I would assume so based on not mentioning it in documentation...)? Or was it just forgotten? I find such value_type quite useful... And since I'm already writing let me note that the mentioned documentation has also another minor issue. In Specification section the sample in_place_factory and typed_in_place_factory should be named in_place_factory1 and typed_in_place_factory1. As this is how they are really named. Or it should be stated explicitly that the actual name is implementation dependent. Adam Badura

1 0

Boostlog and ASIO crash in shared library
by Ray Cheng 17 Sep '14

17 Sep '14

Hi, My project use a simple main using dlopen to "run" the real main function in target shared library. I use both Boost Log and ASIO (for threadpool) in the project. In Linux I found the ASIO got weird behavior with some un-initialized variables and crashing all the time. I am able to trim down my problem into a small program as below: // *** threadpool.cpp *** #include <iostream> #include <memory> #include <boost/asio/io_service.hpp> #include <boost/bind.hpp> #include <boost/thread/thread.hpp> void print() { std::cout << "print" << std::endl; } struct ThreadPool { void start() { boost::asio::io_service ioService; boost::thread_group threadpool; std::unique_ptr<boost::asio::io_service::work> work = std::make_unique<boost::asio::io_service::work>(ioService); threadpool.create_thread(boost::bind(&boost::asio::io_service::run, &ioService)); for (int i=0; i<100; ++i) { ioService.post(&print); } std::cout << "Post all calculation tasks" << std::endl; work.reset(); threadpool.join_all(); ioService.stop(); std::cout << "Finished all calculation tasks" << std::endl; } }; extern "C" void createthreadpool() { ThreadPool pool; pool.start(); } // *** threadpool.cpp *** g++ -O0 -std=gnu++1y -fPIC -c -g -DLinux -DGNU_GCC -I${BOOST_HOME}/include threadpool.cpp g++ -shared -o libthreadpool.so threadpool.o -L${BOOST_HOME}/lib -lboost_system -lboost_date_time -lboost_thread -lboost_filesystem -lboost_regex -lboost_chrono -ldl -lstdc++ // *** main.cpp *** #include <dlfcn.h> #include <iostream> int main(int argc, char** argv) { typedef void* createProcess(); auto service = dlopen("libthreadpool.so", RTLD_LAZY); createProcess* make = reinterpret_cast<createProcess*>(dlsym(service, "createthreadpool")); std::cout << "dlsym" << std::endl; make(); return 0; } // *** main.cpp *** g++ -O0 -std=gnu++1y -fPIC -c -g -DLinux -DGNU_GCC main.cpp // Link against boost_log to simulate my problem (although it is not using any boost_log). This version will crash g++ -o main.bad main.o -L${BOOST_HOME}/lib -lboost_system -lboost_date_time -lboost_thread -lboost_filesystem -lboost_regex -lboost_chrono -lboost_log -ldl -lstdc++ // it works fine If link against libthreadpool, surprising me g++ -o main main.o -L/home/sysmmbuild/dev/boost -lthreadpool -L${BOOST_HOME}/lib -lboost_system -lboost_date_time -lboost_thread -lboost_filesystem -lboost_regex -lboost_chrono -lboost_log -ldl -lstdc++ One more finding is, if compile boost_log without thread support, the main.bad also working fine. I have attached the code and make.sh as make script. Software version Redhat enterprise 6.5 g++ 4.9.1 Boost 1.56.0 Rgds, Ray

2 2