- Boost-users - lists.boost.org

cross-compilation x86_64-w64-ming32: undefined reference to `InterlockedIncrement'
by Frédéric Bron 16 Aug '13

16 Aug '13

Hi, I am trying to cross-compile from linux 64 bits to windows 64 bits. I am using fedora 19 with -std=c++11 and boost 1.53.0 that comes with fedora. The compiler is x86_64-w64-mingw32-g++. The compile flags I use: -O3 -DNDEBUG -DWINDOWS -std=c++11 -Wall -Wextra -pipe -ffloat-store -DBOOST_DISABLE_ASSERTS -DBOOST_FILESYSTEM_VERSION=3 -DBOOST_FILESYSTEM_NO_DEPRECATED -DBOOST_THREAD_USE_LIB and a typical link command: $ x86_64-w64-mingw32-g++ -o foo.exe bar1.o bar2.o -Wl,--enable-auto-import -static -lboost_system-mt -lboost_filesystem-mt -lboost_thread-mt -lboost_chrono-mt -mthreads I then get the following errors (each line multiple times with different .text+0x###): /usr/x86_64-w64-mingw32/sys-root/mingw/lib/../lib/libboost_thread-mt.a(thread.o):(.text+0x286): undefined reference to `InterlockedIncrement' /usr/x86_64-w64-mingw32/sys-root/mingw/lib/../lib/libboost_thread-mt.a(thread.o):(.text+0x3e7): undefined reference to `InterlockedDecrement' /usr/lib64/gcc/x86_64-w64-mingw32/4.8.1/../../../../x86_64-w64-mingw32/bin/ld: /usr/x86_64-w64-mingw32/sys-root/mingw/lib/../lib/libboost_thread-mt.a(thread.o): bad reloc address 0x8 in section `.data' collect2: error: ld returned 1 exit status make: *** [mgw64-release/bin/transpose.exe] Erreur 1 I see that this may come from <boost/detail/interlocked.hpp> but I have no idea what this header is used for. This is what remains after the pre-processor phase (g++ -E): # 1 "/usr/x86_64-w64-mingw32/sys-root/mingw/include/boost/detail/interlocked.hpp" 1 3 4 # 137 "/usr/x86_64-w64-mingw32/sys-root/mingw/include/boost/detail/interlocked.hpp" 3 4 namespace boost { namespace detail { extern "C" long InterlockedIncrement( long volatile * ); extern "C" long InterlockedDecrement( long volatile * ); extern "C" long InterlockedCompareExchange( long volatile *, long, long ); extern "C" long InterlockedExchange( long volatile *, long ); extern "C" long InterlockedExchangeAdd( long volatile *, long ); extern "C" void* InterlockedCompareExchangePointer( void* volatile *, void*, void* ); extern "C" void* InterlockedExchangePointer( void* volatile *, void* ); } } Note that targeting windows 32 bits works allright but the result of the pre-processor phase is slightly different: # 1 "/usr/i686-w64-mingw32/sys-root/mingw/include/boost/detail/interlocked.hpp" 1 3 4 # 137 "/usr/i686-w64-mingw32/sys-root/mingw/include/boost/detail/interlocked.hpp" 3 4 namespace boost { namespace detail { extern "C" __attribute__((dllimport)) long __attribute__((__stdcall__)) InterlockedIncrement( long volatile * ); extern "C" __attribute__((dllimport)) long __attribute__((__stdcall__)) InterlockedDecrement( long volatile * ); extern "C" __attribute__((dllimport)) long __attribute__((__stdcall__)) InterlockedCompareExchange( long volatile *, long, long ); extern "C" __attribute__((dllimport)) long __attribute__((__stdcall__)) InterlockedExchange( long volatile *, long ); extern "C" __attribute__((dllimport)) long __attribute__((__stdcall__)) InterlockedExchangeAdd( long volatile *, long ); } } Frédéric

1 0

Re: [Boost-users] Serialising templates with smart pointers....
by Marcus MacWilliam 16 Aug '13

16 Aug '13

I think my question is how do I add a serialize method to an interface class, which will invoke the serialize method on the derived sub-class. class Interface { public: virtual void aVirtual() = 0; private: friend class boost::serialization::access; template<class Archive> void serialize(Archive & ar, const unsigned int version) { call serialize on derived class } }; class Derived : Interface virtual void aVirtual() { // Do something } private: friend class boost::serialization::access; template<class Archive> void serialize(Archive & ar, const unsigned int version) { ar & m_data; } }; -- Marcus A.T MacWilliam, Senior Software Engineer. 1Spatial Ltd. Tel: +44 (0)1223 420414, ext 2289, Mob: +44 (0)7803 706597. E-Mail: marcus.macwilliam(a)1spatial.com Skype:marcus.macwilliam Tennyson House, 7 Cambridge Business Park, Cambridge, CB4 0WZ. Registered in England No. 4785688, VAT Reg. No. 135539408. Website: http://www.1spatial.com/ E-mail: info(a)1spatial.com

2 1

Re: [Boost-users] compiling error after upgrading from boost 1.33.1 to boost 1.36.0
by Henson Huang 16 Aug '13

16 Aug '13

Hi Igor, I'm using "new" to allocate the object. My code looks like, In hpp file, typedef Std::map<Std::string, int> MyMap; class MyClass { ........... private boost::shared_ptr< MyMap > m_myMap; } In cpp file MyClass::MyClass() { m_ myMap.reset(new MyMap); } MyClass::~MyClass() { m_ myMap.reset(); } Best regards, Hua

1 0

compiling error after upgrading from boost 1.33.1 to boost 1.36.0
by Henson Huang 15 Aug '13

15 Aug '13

Hello, I upgraded the boost library in my workstation from boost 1.33.1 to boost 1.36.0. When I compiled my program, I got following errors, /usr/include/boost/detail/sp_counted_impl.hpp:192: error: template argument 3 is invalid /usr/include/boost/detail/sp_counted_impl.hpp:198: error: 'a' has not been declared /usr/include/boost/detail/sp_counted_impl.hpp: In constructor 'boost::detail::sp_counted_impl_pda<P, D, <template-parameter-1-3> >::sp_counted_impl_pda(P, D, int)': /usr/include/boost/detail/sp_counted_impl.hpp:198: error: 'a' was not declared in this scope /usr/include/boost/detail/sp_counted_impl.hpp: In member function 'virtual void boost::detail::sp_counted_impl_pda<P, D, <template-parameter-1-3> >::destroy()': /usr/include/boost/detail/sp_counted_impl.hpp:209: error: expected nested-name-specifier before 'template' /usr/include/boost/detail/sp_counted_impl.hpp:209: error: expected unqualified-id before 'template' /usr/include/boost/detail/sp_counted_impl.hpp:211: error: 'A2' was not declared in this scope /usr/include/boost/detail/sp_counted_impl.hpp:211: error: expected `;' before 'a2' /usr/include/boost/detail/sp_counted_impl.hpp:214: error: 'a2' was not declared in this scope I'm using gcc 4.1.2. Below is the file sp_counted_impl.hpp line 181 ~ 221, template<class P, class D, class A> class sp_counted_impl_pda: public sp_counted_base { private: P p_; // copy constructor must not throw D d_; // copy constructor must not throw A a_; // copy constructor must not throw sp_counted_impl_pda( sp_counted_impl_pda const & ); sp_counted_impl_pda & operator= ( sp_counted_impl_pda const & ); typedef sp_counted_impl_pda<P, D, A> this_type; //This is the line 192 public: // pre: d( p ) must not throw sp_counted_impl_pda( P p, D d, A a ): p_( p ), d_( d ), a_( a ) { } virtual void dispose() // nothrow { d_( p_ ); } virtual void destroy() // nothrow { typedef typename A::template rebind< this_type >::other A2; A2 a2( a_ ); this->~this_type(); a2.deallocate( this, 1 ); } virtual void * get_deleter( detail::sp_typeinfo const & ti ) { return ti == BOOST_SP_TYPEID( D )? &reinterpret_cast<char&>( d_ ): 0; } }; Please help to see how to resolve these errors. Best regards, Hua

2 1

Re: [Boost-users] Boost-users Digest, Vol 3537, Issue 1
by Rick Aurbach 15 Aug '13

15 Aug '13

Michael, It sounds to me like you're making this altogether too complicated. Don't focus on the software; focus on the science. The size of the "degree" unit is the same for Kelvin and Celsius. (I.e., if the temperature changes by 1 degree K, it changes by exactly 1 degree C.) The only this that is different between Kelvin and Celsius is the zero point. For Kelvin, 0 degrees is absolute zero and there are no negative values of temperature. (Well, that's not quite right, but you'll never encounter the exceptions.) For Celsius, 0 degrees is the freezing point of water at sea level. To convert between Kelvin and Celcius, just add or subtract a constant. That constant has a value of 273.15. That is, you can make a dimensionally-safe conversion from Kelvin to Celcius by subtracting 273.15. Converting from Celsius to Kelvin by adding 273.15. Cheers, Rick Aurbach Aurbach & Associates, Inc. On Aug 15, 2013, at 9:24 AM, boost-users-request(a)lists.boost.org<mailto:boost-users-request@lists.boost.org> wrote: From: Michael Powell <mwpowellhtx(a)gmail.com<mailto:mwpowellhtx@gmail.com>> Subject: [Boost-users] [units] "Simple" conversion from Kelvin to Celsius Date: August 14, 2013 2:39:56 PM CDT To: "boost-users(a)lists.boost.org<mailto:boost-users@lists.boost.org>" <boost-users(a)lists.boost.org<mailto:boost-users@lists.boost.org>> Reply-To: "boost-users(a)lists.boost.org<mailto:boost-users@lists.boost.org>" <boost-users(a)lists.boost.org<mailto:boost-users@lists.boost.org>> Hello, It's been a couple of years since I've done any work with Boost.Units, and I am finding that I want to make a dimensionally-safe conversion from Kelvin to Celsius. I've got an I2C controller that reports in Kelvin, but I want to report it to the end-user in Celsius. The examples I am finding seem to "almost" fit the bill, but I am having a difficult time extrapolating a sane conversion. First, would help if I understood what the Boost.Units SI (my assumed base-system) base-temperature units were. I gather Fahrenheit? Or is it indeed Kelvin? I could be mistaken. Thank you... Regards, Michael Powell

1 0

[boost::log] how to format a record message
by Eric Prud'hommeaux 15 Aug '13

15 Aug '13

I have a project which has a handy stream filter in the formatter which indents newlines in logged messages to some depth à la: I/O [- 2 +] <loadedData> { <s> <p> <o> . <s> <p> <o2> . } </loadedData> (No, I'm not using an XML decorator above, though perhaps I should be.) I've maintained this for several generations of boost::log, e.g. #if (BOOST_VERSION > 104600) // could use (BOOST_LOG_COMPATIBILITY == "1.46") if cpp compared strings sink->set_formatter(&LineIndenter); #else backend->set_formatter(&LineIndenter); #endif but now that boost::log's released (1000s of thanks!!!), I'd like to switch to a channel_severity_filter. I had a custom formatter before in order to fiddle with the rec.message: inline void LineIndenter (std::ostream& strm, logging::record const& rec) { std::stringstream ss; oprfxnstream prfxstr(ss.rdbuf(), depth, ' '); ... stuff with attrs prfxstr << rec.message(); strm << ss.str(); } That member seems to be gone from logging::record and besides, it'd be nicer to use something like: boost::log::add_console_log (std::clog, boost::log::keywords::filter = min_severity || severity >= critical, boost::log::keywords::format = (boost::log::expressions::stream << line_id << ": <" << severity << "> [" << channel << "] " << lineIndenter(boost::log::expressions::smessage) ) ); I haven't figured out the signature for lineIndenter. I have the impression that I'm implementing a decorator like boost/log/expressions/formatters/c_decorator.hpp . So what's an easy way to wrap the attached stream filter (or a subclass constructed with some necessary constants) and pass it to a formatter? -- -ericP

1 0

graph theoretical problem
by Peter Foelsche 15 Aug '13

15 Aug '13

Dear All, does the boost graph library provide a solution to the problem I'm trying to solve: I want to distribute a set of tasks to multiple CPUs/Threads. The tasks have predecessor and successor dependencies. There are no loops. More often than not, a task consists of only very few operations. So if I brute force program this, the resulting code is busy only with scheduling -- the real code does not show up in the profiling results. I think this can be solved by collapsing some tasks into a single task to minimize scheduling/communication. I've got a cost function associated with every task. Is this some well known algorithm which is already available in boost graph? Kind Regards Peter Foelsche

3 5

[bgl] adding nodes recursively
by Littlefield, Tyler 14 Aug '13

14 Aug '13

Hello all: I have a game with multiple rooms, which I would like to map to a graph. Given this, I have a couple of questions. One of my goals is to separate each individual area and have it on it's own graph, then connect the graphs. Is this a possibility? I remember reading about connected graphs for my data structures class, but we didn't get to far with them. My second issue is I will have a vector of rooms, which all have exits. Each individual room will be a vertex with the exits being the edges. I can recursively add all the exits that a room points to, then come back and add the rest of the exits. But there may be two exits pointing to the same room. Is there a quick way that I could insure that rather than creating a new vertex, I could just point the edge to the preexisting vertex? Finally, i have one other question. I plan to create resources based on a bunch of different factors. Basically I need a way to determine the closest resources and perhaps get the shortest path to them, or the weight of all of the edges that would point to the n closest resources that were spawned. I want to use this weight to determine the density or find another way to determine density and spread the resources out across the world. -- Take care, Ty http://tds-solutions.net He that will not reason is a bigot; he that cannot reason is a fool; he that dares not reason is a slave.

1 0

Stability of Boost::Serialization XML Output
by Andreas Neustifter 14 Aug '13

14 Aug '13

Hi List, I have to provide digital signatures of serialized C++ objects, I'm planing on using Boost::Serialization for the XML part. Since XML signatures and XMLDsig in particular are difficult [1] and littered with pitfalls I was wondering if maybe the XML output of Boost::Serialization is stable enough to do a plain byte-oriented PKCSwhatever signature? To summarize: - C++ objects are serialized to XML with Boost::Serialization (XML is used since date has to be as human readable as possible). - Serialized XML has to be digitally signed. - XMLDsig is complicated [1] Question: Does Boost::Serialization with the XML backend produce bytewise the same data every time? I sincerely hope that I made myself clear. Andreas [1] http://www.cs.auckland.ac.nz/~pgut001/pubs/xmlsec.txt

2 2

[msm] exit points & favor compile time policy
by Alexander Mingalev 14 Aug '13

14 Aug '13

Hi, I faced to a strange problem. I'm working with root fsm & child fsm. Child fsm has an exit point called "End". I use vs2012u3, boost 1.54 transition table from root looks like that: struct Waiting : state<> {}; // Child is typedef for back-end struct transition_table : public boost::mpl::vector< ... Row < Waiting, begin, Child >, Row < Child::exit_pt<Child::End>, none, Waiting >, ... > {}; So, when I use F-C-T- policy, the Child fsm is stuck in Child::End state, the root fsm never moves to Waiting again. It doesn't happen with default policy. I noticed, that there is a note in msm's changelog From V2.24 to V2.25 (Boost 1.54) • Bugfix: Exit points broken for the favor_compile_time policy. looks similar to my case. Christophe, can you check that bugfix?

3 3