- Boost - lists.boost.org

Suppressing boost.thread and boost.system symbols in shared library
by Philip Bennefall 02 Aug '13

02 Aug '13

Hi all, I am building a shared library for Linux, for which I would like to export as few symbols as possible (ideally only the functions that my library users actually need to call). When building my project, I have the following options: -fvisibility=hidden -fvisibility-inlines-hidden I then mark all the functions that I am exporting as having default visibility. I link to static versions of boost.thread and boost.system. When I then say: nm -DC libtest.so I get a large number of symbols such as: 0002abe0 V typeinfo name for boost::exception_detail::clone_impl<boost::exceptio n_detail::bad_exception_> 0002ace0 V typeinfo name for boost::exception_detail::bad_exception_ 00029920 V typeinfo name for boost::exception_detail::error_info_injector<boost: :lock_error> 00029b00 V typeinfo name for boost::exception_detail::error_info_injector<boost: :condition_error> 00029860 V typeinfo name for boost::exception_detail::error_info_injector<boost: :thread_resource_error> 0002970f V typeinfo name for boost::thread_exception 00029a7b V typeinfo name for boost::thread_interrupted 00029740 V typeinfo name for boost::thread_resource_error 0002ac40 V typeinfo name for boost::enable_shared_from_this<boost::detail::threa d_data_base> 0002a720 V typeinfo name for boost::detail::thread_data_base 0002ad20 V typeinfo name for boost::detail::sp_counted_impl_p<boost::detail::thr ead_data_base> 0002adc0 V typeinfo name for boost::detail::sp_counted_impl_p<boost::exception_d etail::clone_impl<boost::exception_detail::bad_alloc_> > 0002ad60 V typeinfo name for boost::detail::sp_counted_impl_p<boost::exception_d etail::clone_impl<boost::exception_detail::bad_exception_> > 000296f1 V typeinfo name for boost::system::system_error I see this after having stripped the binary. Is there any way to get rid of these? Do I need to build Boost with some special cxxflags? Kind regards, Philip Bennefall

3 6

[predef] FYI..
by Rene Rivera 01 Aug '13

01 Aug '13

FYI.. to those that care.. I checked in changes to SVN to inject the Predef library into Boost. I say inject because the source is not actually in the SVN repo, it's external and still in github. But is being brought in through their SVN interface. Some testers have already runs the tests for it.. And at some point to be determined with some coordination with the community I will switch the "boost/detail/endian.hpp" header to point to the Predef compatibility equivalent... I suspect at the earliest one more week for that assuming no problems surface. Rene. -- -- -- Grafik - Don't Assume Anything -- Redshift Software, Inc. - http://redshift-software.com -- rrivera/acm.org - grafik/redshift-software.com -- 102708583/icq - grafikrobot/aim - grafikrobot/yahoo

6 10

[predef] A few questions
by Andrey Semashev 01 Aug '13

01 Aug '13

Hi, I have a few questions/nitpicks regarding Boost.Predef: - I see that all headers have extension .h. Why is it not .hpp, like with the rest of Boost? It this intentional? - There is a tab in predef/version_number.h. There are also tabs in tests. - There are __SunOS_5_10, __SunOS_5_11 and similar macros defined on Solaris, which allow to detect OS version. These are not always available but I think SunPro defines them. Could these macros be used to select the proper OS version? - If I'm not mistaken, BOOST_COMP_GNUC and BOOST_COMP_MSVC will get defined not only for the real GCC and MSVC but also for pretenders. Is this intentional? If so, could we have special macros to detect these particular compilers? Boost.Config defines BOOST_MSVC and BOOST_GCC for this purpose. - There is considerable overlap with Boost.Config. Is there any plan to reduce the duplication? Maybe move some things from Boost.Config to Boost.Predef?

3 2

[boost.thread] Trouble building with MinGw 4.7.0
by Philip Bennefall 01 Aug '13

01 Aug '13

Hi all, Up until now I have been developing exclusively with VC++ 2010, and there my project builds just fine and without warnings. However, I am also planning to target Unix so I figured that a first step would be to make sure my project builds with MinGw. I set up my path environment variable to include the location of the MinGw 4.7.0 bin directory, and I built Boost with the following parameters: b2 variant=release link=static runtime-link=static threading=multi toolset=gcc Everything seems to be working fine (stage/lib contains all the .a files that I would expect). However, when I attempt to build my program I get tons of errors like the following: myprog_main.o:myprog_main.cpp:(.text+0x28ef): undefined reference to `__imp___ZN5boost6thread9interruptEv' myprog_main.o:myprog_main.cpp:(.text+0x294f): undefined reference to `__imp___ZNK5boost6thread6get_idEv' myprog_main.o:myprog_main.cpp:(.text+0x295d): undefined reference to `__imp___ZN5boost11this_thread6get_idEv' myprog_main.o:myprog_main.cpp:(.text+0x2973): undefined reference to `__imp___ZN5boost6thread13join_noexceptEv' myprog_main.o:myprog_main.cpp:(.text+0x2981): undefined reference to `__imp___ZN5boost6thread6detachEv' myprog_main.o:myprog_main.cpp:(.text+0x4600): undefined reference to `__imp___ZN5boost11this_thread18interruptible_waitEPvNS_6detail7timeoutE' myprog_main.o:myprog_main.cpp:(.text+0x49c2): undefined reference to `__imp___ZN5boost11this_thread18interruptible_waitEPvNS_6detail7timeoutE' myprog_main.o:myprog_main.cpp:(.text+0x667e): undefined reference to `__imp___ZN5boost11this_thread18interruptible_waitEPvNS_6detail7timeoutE' myprog_main.o:myprog_main.cpp:(.text+0x90a9): undefined reference to `__imp___ZN5boost6thread21start_thread_noexceptEv' myprog_main.o:myprog_main.cpp:(.text+0x9134): undefined reference to `__imp___ZN5boost6thread21start_thread_noexceptEv' myprog_main.o:myprog_main.cpp:(.text+0x917e): undefined reference to `__imp___ZN5boost6thread9interruptEv' myprog_main.o:myprog_main.cpp:(.text$_ZN5boost6detail11thread_dataINS_3_bi6bind_tIvPFvPN5myprog5themeEENS2_5list1INS2_5valueIS6_EEEEEEED1Ev[boost::detail::thread_data<boost::_bi::bind_t<void, void (*)(myprog::theme*), boost::_bi::list1<boost::_bi::value<myprog::theme*> > > >::~thread_data()]+0x7): undefined reference to `__imp___ZN5boost6detail16thread_data_baseD2Ev' myprog_main.o:myprog_main.cpp:(.text$_ZN5boost6detail11thread_dataINS_3_bi6bind_tIvPFvPN5myprog5themeEENS2_5list1INS2_5valueIS6_EEEEEEED0Ev[boost::detail::thread_data<boost::_bi::bind_t<void, void (*)(myprog::theme*), boost::_bi::list1<boost::_bi::value<myprog::theme*> > > >::~thread_data()]+0xe): undefined reference to `__imp___ZN5boost6detail16thread_data_baseD2Ev' myprog_main.o:myprog_main.cpp:(.text$_ZN5boost6threadD1Ev[boost::thread::~thread()]+0x9): undefined reference to `__imp___ZN5boost6thread6detachEv' Etc... I tried to specify -DBOOST_NO_LIB and add all the libraries manually, but no luck. Then I thought that maybe there's a problem with the fact that there are spaces in some of my directory names, so I copied the relevant .a files to the same directory as the program itself but still with no success. At this point, I'm swamped. I have managed to compile MinGw programs in the past that used Boost Thread, so I have no idea why it is not working with this version of Boost. Any suggestions would be highly appreciated. Kind regards, Philip Bennefall

2 5

Re: [boost] [multimethods] Proposal: Open Multi-Methods
by Daryle Walker 01 Aug '13

01 Aug '13

//================================================================= From: Jean-Louis Leroy Sent: ‎Saturday‎, ‎July‎ ‎27‎, ‎2013 ‎8‎:‎26‎ ‎PM I am working on library that pretty much implements what Pirkelbauer, Solodkyy and Stroustrup describe in their paper "Open Multi-Methods for C++" - although there are some divergences. I have posted it on GitHub: https://github.com/jll63/multimethods. Please examine the files in the "examples" directory to get an idea of how to use it. [SNIP] The library is essentially feature complete (if anything ever is). I am now going to work on the documentation, a build system, some clean up and compatibility (I tested it only with g++ 4.7.2 only so far). Is there interest in the Boost community for this stuff? If yes, I will adapt it to the guidelines and submit it for formal review. Of course, suggestions are welcome. //================================================================= Are virtual base classes supported? Maybe “virtual_” should be renamed to “virtualize” or “virtualized,” to avoid having a trailing underscore in a non-member name. A good way to document is to write a Doxygen comment as you write your class, function, or macro prototype. It makes you have to think about what the public side of your code has to do. But looking at your “next.cpp” example, you should start with documenting the “selector,” “virtual_,” “MM_CLASS,” “MM_INIT,” “MULTIMETHOD,” “BEGIN_METHOD,” “END_METHOD,” and “initialize” items. Anything in your implementation classes that don’t have to be seen by public code can be make private. Non-public types can go in a detail namespace. Daryle W.

3 2

[boost thread] Catching interrupted exception?
by Philip Bennefall 31 Jul '13

31 Jul '13

Hi all, I have a quick question regarding the interruption of Boost threads. I am making fairly heavy use of the interrupt method call of the thread class, and I am wondering whether I should be catching the boost::thread_interrupted exception from within my thread function? I currently do not and my program runs fine, but I am wondering whether this is defined behavior or whether I should always be catching the exception and returning from the function inside the handler? Kind regards, Philip Bennefall

2 2

Re: [boost] [asio] Randomly having problems reading from the socket
by Marat Abrarov 30 Jul '13

30 Jul '13

> Posted a Bug Report here: Ticket #8937 > <https://svn.boost.org/trac/boost/ticket/8937> Seems like a bug/feature of Windows 7/Vista (or a bug/feature of Windows socket subsystem) - something in background/foreground process scheduling or a memory management issue. Nevertheless it is interesting to know the Asio author's point of view. It will be more clear if the topic starter rewrites his (simple) examples to use WinSock directly rather than through the Asio (use the same time measurement - by means of boost::posix_time). I think that such rewritten examples will show the same behavior. Regards, Marat Abrarov.

1 0

[preprocessor] ARRAY_REPLACE bug?
by Mostafa 30 Jul '13

30 Jul '13

On Windows 7/Cygwin, using boost 1.53, preprocessing the following snippet: pp_array.cc ----------- #include <boost/preprocessor.hpp> //#define BOOST_PP_TUPLE_REM_0() // nothing #define ARRAY (3, (a, x, c)) BOOST_PP_ARRAY_REPLACE(ARRAY, 1, b) via: g++ -E -P -I /... pp_array.cc gives the following incorrect result: (3, (BOOST_PP_TUPLE_REM_0 () a, b , c)) Uncommenting the pound define in the snippet gives the correct result for the preprocessor. Is this a bug?

3 5

[type_erasure] move
by Christophe Henry 29 Jul '13

29 Jul '13

Hi Steven, I'm looking for a crash I get in my own code with g++4.8.1 (4.5 - 4.7 are ok). Maybe you can help me: - there is a copy_constructible concept, but no moveable. Is it enough to add one on the same model? It looks like you std::forward so I'd say yes. Correct? template<class T = _self> struct move_constructible : ::boost::mpl::vector<constructible<T(T&&)>, destructible<T> > {}; - If I std::move a copy_constructible any type by accident, what do I get? A copy? A crash? Thanks, Christophe

2 2

[asio] Randomly having problems reading from the socket
by Torben Wiggerich 29 Jul '13

29 Jul '13

Hi all, I have some problems using the boost.asio.read for a TCP-socket. So far it works fine, but it randomly stops the transfer for at least 5 seconds. To resolve the error I used a simple synchronous server and client from the boost examples and modified it too send a larger amount of data. (307400 Bytes; this size comes from the program I'm trying to get run). On the client side I also changed the reading from the socket to read the complete data with boost.asio.read and transfer_all as argument. Both, server and client, measure the time for sending and reading to the socket and will give an error if it needs more than 1 second. Here is the sourcecode for the server: #include <iostream> #include <string> #include <boost/asio.hpp> #include <boost/date_time/posix_time/posix_time.hpp> using boost::asio::ip::tcp; using namespace boost::posix_time; int main() { try { boost::asio::io_service io_service; tcp::acceptor acceptor(io_service, tcp::endpoint(tcp::v4(), 8002)); tcp::socket socket(io_service); acceptor.accept(socket); unsigned char * array = new unsigned char[307400]; for (;;) { boost::system::error_code ignored_error; ptime startTime = microsec_clock::universal_time(); boost::asio::write(socket, boost::asio::buffer( array, 307400 ), boost::asio::transfer_all(), ignored_error); time_duration neededTime = microsec_clock::universal_time() - startTime; if ( neededTime.total_milliseconds() > 1000 ) { std::cout << "ERROR needed too much time: " << neededTime.total_milliseconds() << "ms" << std::endl; } } } catch (std::exception& e) { std::cerr << e.what() << std::endl; } return 0; } My client sourcecode: #include <iostream> #include <boost/array.hpp> #include <boost/asio.hpp> #include <boost/date_time/posix_time/posix_time.hpp> using boost::asio::ip::tcp; using namespace boost::posix_time; int main(int argc, char* argv[]) { try { if (argc != 2) { std::cerr << "Usage: client <host>" << std::endl; return 1; } boost::asio::io_service io_service; boost::asio::ip::tcp::endpoint endpoint( boost::asio::ip::address::from_string( argv[1] ), 8002 ); tcp::socket socket(io_service); boost::system::error_code error = boost::asio::error::host_not_found; socket.connect( endpoint, error ); if (error) throw boost::system::system_error(error); int counter = 0; unsigned char* array = new unsigned char[307400]; for (;;) { boost::system::error_code error; ptime startTime = microsec_clock::universal_time(); size_t len = boost::asio::read( socket, boost::asio::buffer( array, 307400 ), boost::asio::transfer_all(), error ); time_duration neededTime = microsec_clock::universal_time() - startTime; if ( error ) { printf( "body read error %s\n", error.message().c_str() ); } if ( counter % 5000 == 0 ) { std::cout << "passed: " << counter << "\r"; } ++counter; if ( neededTime.total_milliseconds() > 1000 ) { std::cout << "ERROR needed too much time: " << neededTime.total_milliseconds() << "ms" << std::endl; } } } catch (std::exception& e) { std::cerr << e.what() << std::endl; } return 0; } To reproduce the error just let it run for a while. It can take up to 30 minutes or even more... After a while it will produce the line: "ERROR needed too much time: 5007ms". Maybe with slightly different times, but all around 5 seconds. I tested it on different machines with Windows Vista and Seven. So am I'm doing sth. wrong? Is the reading from the socket correct? I'm using boost version 1.44. Kind regards Torben Wiggerich

6 12