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

Signals - no linking
by michael graves ii 09 Aug '02

09 Aug '02

hey, when trying to run a program using boost::signal i am getting a lot of unresolved errors. [output window] --------------------Configuration: my_dlls - Win32 Release-------------------- Linking... delegate_main.obj : error LNK2001: unresolved external symbol "public: __thiscall boost::signals::detail::signal_base::~signal_base(void)" (??1signal_base@detail@signals@boost@@QAE@XZ) delegate_main.obj : error LNK2001: unresolved external symbol "protected: __thiscall boost::signals::trackable::~trackable(void)" (??1trackable@signals@boost@@IAE@XZ) delegate_main.obj : error LNK2001: unresolved external symbol "public: void __thiscall boost::signals::connection::disconnect(void)const " (?disconnect@connection@signals@boost@@QBEXXZ) delegate_main.obj : error LNK2001: unresolved external symbol "protected: void __thiscall boost::signals::detail::slot_base::create_connection(void)" (?create_connection@slot_base@detail@signals@boost@@IAEXXZ) delegate_main.obj : error LNK2001: unresolved external symbol "public: __thiscall boost::signals::detail::signal_base_impl::signal_base_impl(class boost::function2<bool,class boost::any,class boost::any,struct boost::empty_function_policy,struct boo st::empty_function_mixin,int> const &)" (??0signal_base_impl@detail@signals@boost@@QAE@ABV?$function2@_NVany@boost@@ V12@Uempty_function_policy@2@Uempty_function_mixin@2@H@3@@Z) delegate_main.obj : error LNK2001: unresolved external symbol "public: class boost::signals::connection __thiscall boost::signals::detail::signal_base_impl::connect_slot(class boost::any const &,class boost::any const &,class std::vector<class boost ::signals::trackable const *,class std::allocator<class boost::signals::trackable const *> > const &)" (?connect_slot@signal_base_impl@detail@signals@boost@@QAE?AVconnection@34@AB Vany@4@0ABV?$vector@PBVtrackable@signals@boost@@V?$allocator@PBVtracka ble@signals@boost@@@std@@@std@@@Z) delegate_main.obj : error LNK2001: unresolved external symbol "public: __thiscall boost::signals::detail::call_notification::~call_notification(void)" (??1call_notification@detail@signals@boost@@QAE@XZ) delegate_main.obj : error LNK2001: unresolved external symbol "public: __thiscall boost::signals::detail::call_notification::call_notification(class boost::shared_ptr<class boost::signals::detail::signal_base_impl> const &)" (??0call_notification@de tail@signals@boost@@QAE@ABV?$shared_ptr@Vsignal_base_impl@detail@signals@boo st@@@3@@Z) delegate_main.obj : error LNK2001: unresolved external symbol "public: __thiscall boost::signals::detail::signal_base_impl::~signal_base_impl(void)" (??1signal_base_impl@detail@signals@boost@@QAE@XZ) Release/my_dlls.exe : fatal error LNK1120: 9 unresolved externals Error executing link.exe. my_dlls.exe - 10 error(s), 0 warning(s) [/output window] so in semi-readable for these are the unresolved symbols: signal_base::~signal_base(void) trackable::~trackable(void) void connection::disconnect(void)const void slot_base::create_connection(void) signal_base_impl::signal_base_impl(class boost::function2<bool,class boost::any,class boost::any,struct boost::empty_function_policy,struct boost::empty_function_mixin,int> const &) connection signal_base_impl::connect_slot(class boost::any const &,class boost::any const &,class std::vector<class boost::signals::trackable const *,class std::allocator<class boost::signals::trackable const *> >const &) call_notification::~call_notification(void) call_notification::call_notification(class boost::shared_ptr<class boost::signals::detail::signal_base_impl> const &) signal_base_impl::~signal_base_impl(void) i downloaded the signals library out of the CVS Repository. i see that some people have gotten signals to work so i guess i'm missing something=( i like this lib a lot (and it saves me from coding it=), but i would have went with a delegate/callback naming scheme instead of signal/slot. i guess that's what find and replace is for=). michael graves ii

3 6

No time/date class in boost?
by Sean Shubin 09 Aug '02

09 Aug '02

I would have expected to find a time/date class in boost. Was it determined to be unnecessary, or is it simply that no one has gotten around to implementing one?

4 3

Regex issue on WIN32
by Rob Reece 09 Aug '02

09 Aug '02

I'm trying to get the regex lib (boost 1.28.0) built on win32 (MSVC v6, stlport 4.5.3). The standard build that is provided with regex appears to work fine and I get the full compliment of regex libraries (debug, static/dynamic, etc.). However, when I attempt to build the credit_card_example file, the link fails with unresolved external symbols. I then added the 13 regex source files to a msdev project along with the credit_card_example source file and linked everything statically (using BOOST_REGEX_STATIC_LINK or whatever it is). That worked fine. So this leads me to believe that something is going on with BOOST_REGEX_DECL macro. I put some #pragma message(...) statements into some headers and determined that the macro is being defined correctly. Is it possible that this macro is missing on some functions/classes or is it more likely a build option or macro issue? The first unresolved symbol I have is boost::re_detail::match_results_base<...>::m_free(...) As I was looking around for that symbol, I discovered in regex.hpp that match_results_base did not have BOOST_REGEX_DECL macro on the class definition. I added it and rebuilt and I ended up with more unresolved symbols. So obviously that isn't the direction to go. I have, however, not been able to find the actual definition of m_free yet (still looking). Thanks -Rob

2 1

Graph: Generic access to vertex/edge properties
by Tarjei Knapstad 09 Aug '02

09 Aug '02

Hi. I'm currently trying to implement a generic pattern matching (aka subgraph isomorphism) algorithm for use with the Boost Graph Library. My first-attempt function signature looks like this (comments included for further description of what I'm attempting) //! Seeks to find matches of the graph \a g2 in the graph \a g1. /*! This is called "graph pattern matching", or "subgraph homomorphism/isomorphism". The results are returned in a std::list of std::maps where the maps' key is the node_descriptor in \a g2 and the values are the corresponding node_descriptors in \a g1. The size of the returned list represents the number of matches. If \a properties is true it means that all vertex and edge properties must be equal in both structures for a match to be found. Otherwise, only the graph structure itself will be checked. \sa positioned_pattern_match */ template <class Graph> std::list< std::map<typename graph_traits<Graph>::vertex_descriptor, typename graph_traits<Graph>::vertex_descriptor> > pattern_match(const Graph& g1, const Graph& g2, bool properties = true); For the case where the 'properties' switch is false I have everything I need, but if 'properties' is true I'm a bit unsure on how to proceed. The problem, as I see it, is to do a generic consecutive access of all the properties of the current vertex/edge being checked and see if they're equal in both 'g1' and 'g2'. Could anyone provide me a solution/clue on how to solve this problem in a generic way so that the algorithm can be used for any graph? Regards, -- Tarjei Knapstad

3 2

VC7 build error without #define BOOST_DISABLE_WIN32
by Martin Schuerch 08 Aug '02

08 Aug '02

Hi I have an application with MFC / ATL and also some files compiled with the /clr option. Throuout the application boost is used a lot. All builds well as long as #define BOOST_DISABLE_WIN32 is set in the user.hpp file. Without this option a) the build hangs during registration after linking or b) link error due to differently defined custom attributes (in a file using /clr option) Did anybody had similar problems already? Regards Martin

1 0

iterator_adaptor and operator-> for input iterators
by Beman Dawes 07 Aug '02

07 Aug '02

I wrote an iterator to do recursive directory iteration. The heart of it is: template< typename Predicate > class recursive_directory_iterator : public boost::iterator_adaptor< boost::shared_ptr< detail::rdi_imp< Predicate > >, detail::rdi_policies, const boost::filesystem::path, const boost::filesystem::path &, const boost::filesystem::path *, std::input_iterator_tag, std::ptrdiff_t > { ... }; Works fine as long as you stay away from operator->. For example, fs::recursive_directory_iterator<pred_type> rdi( dir , dummy ); fs::recursive_directory_iterator<pred_type> rdi_end; for ( ; rdi != rdi_end; ++rdi ) { // std::cout << rdi->generic_path() << std::endl; std::cout << (*rdi).generic_path() << std::endl; } works as expected. But if you uncomment the commented line, the program fails in the call to operator->. What "fails" means depends on the execution environment; whatever is happening seems to confuse the debugger, stepping doesn't shed much light on the problem. Same symptoms with both VC++7 and Intel 6.0. I suppose I'm doing something stupid, but don't have a clue as to what. I notice in iterator_adaptor_test.cpp that operator-> is not actually exercised. Is it possible there is a bug in it for input iterators? I've not had similar problems with iterator_adaptor for bidirectional iterators. Any help appreciated, --Beman

1 0

Newbie and his template for comments.
by Kamil Burzynski 07 Aug '02

07 Aug '02

Hello. I'm a C++ developer, who is learning STL recently, and loving it :) I've encountered Boost library, which seems to extending STL funcationality a lot :) I've also created a sampel template for usage in my company, and I would like to get comments about it: is this code ok, exception safe, somebody did it better things like that. In my projects I'm often using structures like map<T1, T2 *>, e.g. map<int, Point *> or map<string, Person*> or something like this. So, often I want to do something like that: map<int, Point *> my_map; map<int, Point *>::iterator it; /* .. */ for( it=my_map.begin();it!=my_map.end();++it ) { it->second->Reset(); //sets x,y to 0,0 } example may be stupid, but shows what I'm trying to explain. To improve it I've created a template (given below), so I ca use it like that: for_each( my_map.begin(), my_map.end(), map_fun( &Point::Reset ) ); Which i think is much better. My template is as follows (style is kept from GCC includes, so it may be not as readable as it could be :) template <class _Ret, class _Tp> class map_fun_t : public unary_function<_Tp*,_Ret> { public: explicit map_fun_t(_Ret (_Tp::*__pf)()) : _M_f(__pf) {} template< class _Mp> _Ret operator()(_Mp &__p) const { return ((__p.second)->*_M_f)(); } private: _Ret (_Tp::*_M_f)(); }; template <class _Ret, class _Tp> inline map_fun_t<_Ret,_Tp> map_fun(_Ret (_Tp::*__f)()) { return map_fun_t<_Ret,_Tp>(__f); } What do you think about it? Has it any flaws? Somebody done it already? Is it usable for anyone except me? :) Thanks in advance. -- Best regards from Kamil Burzynski Senior Design Engineer Advanced Digital Broadcast Poland, LTD. - - "Yes, I'm criminal. My crime is that of curiosity."

4 6

Regex++ link problem with VC6 when building with _UNICODE defined
by Edward Diener 07 Aug '02

07 Aug '02

If I build boost 1.28 regex with _UNICODE defined and then attempt to link the static mssd.lib version with a _UNICODE application using the library, I get: boost_regex_vc6_mssd.lib(w32_regex_traits.obj) : error LNK2005: _iswspace already defined in libcmtd.lib(_wctype.obj) boost_regex_vc6_mssd.lib(c_regex_traits.obj) : error LNK2005: _iswalpha already defined in libcmtd.lib(_wctype.obj) boost_regex_vc6_mssd.lib(c_regex_traits.obj) : error LNK2005: _iswupper already defined in libcmtd.lib(_wctype.obj) boost_regex_vc6_mssd.lib(c_regex_traits.obj) : error LNK2005: _iswlower already defined in libcmtd.lib(_wctype.obj) boost_regex_vc6_mssd.lib(c_regex_traits.obj) : error LNK2005: _iswdigit already defined in libcmtd.lib(_wctype.obj) boost_regex_vc6_mssd.lib(c_regex_traits.obj) : error LNK2005: _iswxdigit already defined in libcmtd.lib(_wctype.obj) boost_regex_vc6_mssd.lib(c_regex_traits.obj) : error LNK2005: _iswspace already defined in libcmtd.lib(_wctype.obj) boost_regex_vc6_mssd.lib(c_regex_traits.obj) : error LNK2005: _iswpunct already defined in libcmtd.lib(_wctype.obj) boost_regex_vc6_mssd.lib(c_regex_traits.obj) : error LNK2005: _iswcntrl already defined in libcmtd.lib(_wctype.obj) Debug/TestRegularExpressionGUIStatic.exe : fatal error LNK1169: one or more multiply defined symbols found Error executing link.exe. This only occurs in the debug multi-threaded static library version. Has anyone else seen this or know why it might occur ? If I change the build options for this library from /Od /Gz to /O2, everything works fine and no linker errors occur.

2 1

[BGL] Common children for vertices in DAG
by Erik Arner 07 Aug '02

07 Aug '02

Is there a fast way, with or without preprocessing of the graph, for checking if two non-adjacent vertices in a DAG have one or more common children? Any pointers to components of the BGL that implement, or can be used to implement, this task would be much appreciated. Thanks, Erik

2 4

Tuple from cons list
by j1k1cki 07 Aug '02

07 Aug '02

Hi, Is there a simple way of obtaining a tuple type out of cons list? It also seems natural to build tuple type from two MPL iterators. Is such code available anywhere already? Best regards

1 0