- Boost-users - lists.boost.org

Re: [Boost-users] (de)serializing map<const A*>
by Cheok Yan Cheng 03 Aug '06

03 Aug '06

Hi Robert, I am using VC++ 2003 I must say, the following workaround just work :) :) :) void load(Archive & ar, const unsigned int version) { void * xptr; xptr = & fiducialPairSearchRegionMap; map<FiducialPair *, vector<RectPair> > *mptr; mptr = static_cast<map<FiducialPair *, vector<RectPair> > >(xptr); ar >> *mptr; } Btw, may I know what is the reason behind of using static_cast instead of reinterpret_cast (reinterpert_cast can be compiled too)? Is it by following a thumb of rule, "use static_cast if you can and use reinterpret_cast if you have to"? Thank you very much! yccheok

2 1

boost::bind for returning a reference?
by Paul 03 Aug '06

03 Aug '06

Hi all, I understand why boost::bind() must return a const-reference or a value, but what I don't understand is, why must I do boost::bind<R&>(...) if I want it to return by reference? Why must I (the user) deduce the return type myself? Can't we have a boost::bind_return_reference(...) function that is the equivalent of boost::bind<R&>(...), so I don't have to do all the hard result_of type deduction manually? cheers Paul

2 3

Boost.Signals doc_view example application question?
by Allen Gooch 02 Aug '06

02 Aug '06

I have a quick question regarding connecting a bound member function. In the doc_view example application signals are defined as requiring a function taking a bool and returning nothing. class Document { public: typedef boost::signal< void (bool) > signal_t; // ... void append(const char* s) { m_text += s; m_sig(true); } // ... }; The View abstract base class requires concrete bases to provide a concrete implementation of the refresh() member function matching the slot signature required. class View { public: virtual void refresh(bool) const = 0; // ... }; Here's my problem. If I change the signal to take no args using the preferred or portable syntax I get compilation errors: class Document { public: typedef boost::signal< void () > signal_t; // ... void append(const char* s) { m_text += s; m_sig(true); } // ... }; class View { public: virtual void refresh() const = 0; // ... }; Compilation output (msvc-7.1): Compiling... SignalTest.cpp c:\Boost\include\boost-1_33_1\boost\bind.hpp(63) : error C2825: 'F::result_type': cannot form a qualified name c:\Boost\include\boost-1_33_1\boost\bind\bind_template.hpp(15) : see reference to class template instantiation 'boost::_bi::result_traits<R,F>' being compiled with [ R=boost::_bi::unspecified, F=void (__thiscall View::* )(void) const ] c:\development\SignalTest\SignalTest.cpp(49) : see reference to class template instantiation 'boost::_bi::bind_t<R,F,L>' being compiled with [ R=boost::_bi::unspecified, F=void (__thiscall View::* )(void) const, L=boost::_bi::list2<boost::_bi::list_av_2<View *,boost::arg<1>>::B1,boost::_bi::list_av_2<View *,boost::arg<1>>::B2> ] c:\Boost\include\boost-1_33_1\boost\bind.hpp(63) : error C2039: 'result_type' : is not a member of 'operator``global namespace''' c:\Boost\include\boost-1_33_1\boost\bind.hpp(63) : error C2146: syntax error : missing ';' before identifier 'type' c:\Boost\include\boost-1_33_1\boost\bind.hpp(63) : error C2955: 'boost::_bi::type' : use of class template requires template argument list c:\Boost\include\boost-1_33_1\boost\bind.hpp(112) : see declaration of 'boost::_bi::type' c:\Boost\include\boost-1_33_1\boost\bind.hpp(63) : fatal error C1903: unable to recover from previous error(s); stopping compilation Observed with boost_1_33_1 release on cygwin/gcc-3.4.4 and msvc-7.1platforms. Any ideas? Many thanks in advance, -allen

1 0

boost::filesystem::ifstream and .eof()
by pktao＠uwm.edu 02 Aug '06

02 Aug '06

Dear Boost users --I have a simple question about how to test for eof. I usually use: std::ifstream inputDataStream("path to file"); and test for EOF by: inputDataStream.eof() I tried the same thing using: boost::filesystem::basic_ifstream<char> inputDataStream("path to file", std::ios_base::in); and test for EOF by: inputDataStream.eof() but its not working. Am I using the .eof() correctly in the boost version of basic_ifstream? Here is the broken code using the boost version of basic_ifstream: int main() { boost::filesystem::path dataPath("/tao/data"); if ( !exists(dataPath) ) { std::cerr << "Path does not exist!"; return EXIT_FAILURE; }; for (boost::filesystem::directory_iterator fileIterator(dataPath); fileIterator != boost::filesystem::directory_iterator(); fileIterator++) { const boost::filesystem::path InputFileName = dataPath/fileIterator -> leaf(); boost::filesystem::basic_ifstream<char> inputDataStream(InputFileName, std::ios_base::in); int i = 1; int previousIterator = 0; trialOnsetWrite(); while ( !inputDataStream.eof() ) { if ( (previousIterator + 3) == i) { immobileWrite(i); previousIterator = previousIterator + 3; } else { dataStructureWrite(i); }; i++; }; trialOffsetWrite(i); }; return (0); }; Very appreciative, Patrick

2 1

Lack of release() in scoped_ptr and scoped_array
by Milutin Jovanovic 02 Aug '06

02 Aug '06

I finally decided to switch to boost from my own implementation of some helper classes, and I noticed that scoped_ptr and scoped_array in boost lack the release method(). I had to fall back to std::auto_ptr. However, this seems like such a basic and fundamental problem that I have started to fear that I am missing something obvious... The usage I got used to was something like this: { scoped_ptr<A> temp( new A() ); verify( temp.get() ); // might throw store_in_a_very_special_place( temp.get() ); // might throw; and I should keep ownership if thrown temp.release(); } I though this was good practice to make code exception safe. Sure I could do it with try-catch but that gets messy if I allocate multiple instances etc. So the question is: 1) is there something wrong with the concept of release()? 2) if release() is not evil, is there any other reason why is it not in boost::scoped_*? Regards, Milutin Jovanovic

3 2

boost::mpl::if_
by Noel Yap 02 Aug '06

02 Aug '06

I'd like to do something like: typedef typename boost::mpl::if_< typename base::is_container<T>::type, typename T::element_type, T>::type ElementType_; such that T may be a primitive type. The above doesn't work since, for example, there's no such thing as int::element_type. Is there a simple way (eg don't have to do template specialization for all container types) to do the above? Thanks, Noel

2 1

enable_if problem
by Imre 02 Aug '06

02 Aug '06

Hi I have some non-compiling code, and I'd like to know if it's a compiler bug, or something I'm doing wrong. In the following code (it's a simplified version of the real code where I met the problem) the call to F<int> instantiates S<int>. I think the compiler should choose the specialized version of S<>, but it seems to use the primary template instead, leading to an invalid non-lvalue int -> int& conversion. Strangely, if I explicitly instantiate S<int> (uncomment the commented line), then it compiles fine. I also tried implementing enable_if and is_fundamental myself, and came to the exact same results. Also tested that version (with the custom enable_if and is_fundamental) with the online Comeau compiler, and it compiled fine. So, I'd like to know whether this stuff should work according to the standard, and if not, why; and if anyone could suggest a workaround that works with VC++ 7.1. Thanks, Imre Here's the code: #include <boost/type_traits.hpp> #include <boost/type_traits/is_fundamental.hpp> #include <boost/utility/enable_if.hpp> template <typename T, typename Enable = void> struct S { typedef T& type; }; template <typename T> struct S<T, typename boost::enable_if_c<boost::is_fundamental<T>::value>::type> { typedef T type; }; //template struct S<int>; // Compiles fine with VC if this line is uncommented template <typename T> typename S<T>::type F() { return 0; } int main(int argc, char* argv[]) { F<int>(); return 0; }

1 0

regex_constants::perl
by Winson Yung 02 Aug '06

02 Aug '06

Hello, I saw the following thread talking about how to set regex to work in perl extended mode: http://groups.google.com/group/boost-list/browse_thread/thread/4d5588fb7021… I am using 1.28.0 and has coded something like this already: boost::RegEx expr(is10k ? keywords_10k[i].m_Key : keywords_10q[i].m_Key, true); // Pass callback, strings to RegEx library unsigned int result = expr.Grep(cb, pBuf); question is how do I set this boost::regex_constants::perl like the follow example: boost::regex rxBlah("([a-z](?:_?[a-z0-9])*)_dvs", boost::regex_constants::perl | boost::regex_constants::icase); /Winson

2 3

replace_all_regex on c-string help
by llwaeva＠21cn.com 02 Aug '06

02 Aug '06

Hi there, I am using in-place replacement with replace_all_regex replace_all_regex( source, boost::regex("(%f)"), "(?1 xxx)", format_all); where source is of type std::string. But for some reason, I need to perform replacement on a C-string (char *), is replace_all_regex safe to do so? Will replace_all_regex handle memory issue for us? Thanks.

3 2

Re: [Boost-users] replace_all_regex on c-string help
by llwaeva＠21cn.com 02 Aug '06

02 Aug '06

>> But for some reason, I need to perform replacement on a >> C-string (char *), is replace_all_regex safe to do so? Will replace_all_regex handle memory >> issue for us? >Of course this would be impossible. Boost.Regex can't know how the >memory was allocated (e.g. malloc or new or some custom allocator), or >even if it *was* allocated. >Why do you need to operate on raw char*'s? Why can't you operate on >an std::string? I got the string (char*s), which is too big (>10M), from an external library, it's too slow to convert the raw string to std::string before it is handled by replace_all_regex

2 1