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

Boost.python-Inner class
by pankaj jain 12 May '05

12 May '05

Hi , I have C++ code like class A { class B; --- --- } can any body help me in how I can map inner class for building python module. Thanks __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

2 1

Can't get "write" to work when using IOStreams with custom streambuf
by Eyal Susser 11 May '05

11 May '05

Hi, Boost is a really neat idea, I wish I had less trouble using it (so embarrasing...). Anyway, here's a streambuf class I wrote using IOStream : class CrIPCStreamOutput : public sink { public: CrIPCStreamOutput(RU8 * pio_pInputStorage , RS32 & pio_nStorageLen) : m_pBuffer(pio_pInputStorage), m_nBufferLen(pio_nStorageLen) { } virtual ~CrIPCStreamOutput() { } void write(const char * s , std::streamsize n) { if(!m_pBuffer) { m_nBufferLen = -1; return; } memcpy(m_pBuffer , s , n); m_nBufferLen = n; return; } private: unsigned char *& m_pBuffer; int & m_nBufferLen; }; In my main, I do: int main() { unsigned char p[256] = {0}; int n = sizeof(p); streambuf_facade<CrIPCStreamOutput> sbo(CrIPCStreamOutput(p , n)); //stream_facade<CrIPCStreamOutput> stro(CrIPCStreamOutput(p , n)); iostream stro(&sbo); stro.write("Hello?" , 6); cout << p << endl; return 0; } But the ouput from cout is an empty string, although when debugging I can see m_pBuffer being copied to. How can I get this to work? The motivation for the whole thing is serialization: I want to serialize a class using the boos.serialization lib, for which I need a stream, which has to have certain properties (not shown here). I need to have the archiver write to a specific buffer. IOStreams should do it (I got the reading side working fine), but how? Thanks.

2 4

Fwd: Re: [Boost-users] Re: Re: Re: Re: Boost serializationanddependencynightmares...
by kauai 11 May '05

11 May '05

Hi Robert et all, Ok details. Now that I am able to compile/link/run using that macro Jeff provided I can move on. Quick summary, for legacy reasons we have a structure called ByteArray passed around everywhere. It is simply a public class like this: class ByteArray { public: unsigned char *pArray; int length; int offset; } On our receiving side we have a piece of code like this: // The function argument(s) have to be like this unfortunately. Also pArray is serialized by the sender with code like this but using binary_oarchive OrderedCollection* OrderedCollection::GetOrderedCollection(ByteArray *pArray) // this is a static method, pArray points to valid heap memory { OrderedCollection *pCollection = NULL; std::string astring ((char*) (pArray->pData), pArray->length + pArray->offset); std::istringstream iss(astring); boost::archive::binary_iarchive ia(iss); // exception here in Windows not Linux, below is the reason and call stack is << BOOST_SERIALIZATION_NVP(pCollection); return pCollection; } Again this works in Linux, and I've run it through Valgrind (a linux tool) with no errors or leaks. If there is a better way to construct a binary input archive I'd like to know, im new to stl classes. I get a std::bad_alloc exception, here is the call stack: cmnTest2.exe!boost::archive::basic_binary_iprimitive<boost::archive::binary_iarchive,std::basic_istream<char,std::char_traits<char> > >::load(std::basic_string<char,std::char_traits<char>,std::allocator<char> > & s={...}) Line 93 C++ > cmnTest2.exe!boost::archive::load_access::load_primitive<boost::archive::binary_iarchive,std::basic_string<char,std::char_traits<char>,std::allocator<char> > >(boost::archive::binary_iarchive & ar={...}, std::basic_string<char,std::char_traits<char>,std::allocator<char> > & t={...}) Line 96 C++ cmnTest2.exe!boost::archive::detail::load_non_pointer_type<boost::archive::binary_iarchive,std::basic_string<char,std::char_traits<char>,std::allocator<char> > >::load_primitive::invoke(boost::archive::binary_iarchive & ar={...}, std::basic_string<char,std::char_traits<char>,std::allocator<char> > & t={...}) Line 308 + 0xd C++ cmnTest2.exe!boost::archive::detail::load_non_pointer_type<boost::archive::binary_iarchive,std::basic_string<char,std::char_traits<char>,std::allocator<char> > >::invoke(boost::archive::binary_iarchive & ar={...}, std::basic_string<char,std::char_traits<char>,std::allocator<char> > & t={...}) Line 366 + 0xd C++ cmnTest2.exe!boost::archive::load<boost::archive::binary_iarchive,std::basic_string<char,std::char_traits<char>,std::allocator<char> > >(boost::archive::binary_iarchive & ar={...}, std::basic_string<char,std::char_traits<char>,std::allocator<char> > & t={...}) Line 529 + 0xd C++ cmnTest2.exe!boost::archive::basic_binary_iarchive<boost::archive::binary_iarchive>::load_override<std::basic_string<char,std::char_traits<char>,std::allocator<char> > >() Line 66 + 0x15 C++ cmnTest2.exe!boost::archive::binary_iarchive_impl<boost::archive::binary_iarchive>::load_override<std::basic_string<char,std::char_traits<char>,std::allocator<char> > >() Line 46 C++ cmnTest2.exe!boost::archive::detail::interface_iarchive<boost::archive::binary_iarchive>::operator>><std::basic_string<char,std::char_traits<char>,std::allocator<char> > >(std::basic_string<char,std::char_traits<char>,std::allocator<char> > & t={...}) Line 84 C++ cmnTest2.exe!boost::archive::detail::common_iarchive<boost::archive::binary_iarchive>::init() Line 66 C++ cmnTest2.exe!boost::archive::binary_iarchive_impl<boost::archive::binary_iarchive>::binary_iarchive_impl<boost::archive::binary_iarchive>(std::basic_istream<char,std::char_traits<char> > & is={...}, unsigned int flags=0) Line 60 C++ cmnTest2.exe!boost::archive::binary_iarchive::binary_iarchive(std::basic_istream<char,std::char_traits<char> > & is={...}, unsigned int flags=0) Line 78 + 0x1e C++ cmnTest2.exe!OrderedCollection::GetOrderedCollection(ByteArray * pArray=0x0034c620) Line 111 + 0x14 C++ Thanks, Kauai --- kauai <kauai_diver(a)yahoo.com> wrote: > Date: Wed, 11 May 2005 10:15:34 -0700 (PDT) > From: kauai <kauai_diver(a)yahoo.com> > Subject: Re: [Boost-users] Re: Re: Re: Re: Boost > serializationanddependencynightmares... > To: boost-users(a)lists.boost.org > > > Hi Robert, Jeff, et all, > > Thank you Jeff, using that macro helped, I can now build > with the Microsoft compiler. Now getting some sort of > exception in my serialization code. > > I am looking into that today, have to setup my visual > studio and debugger :) And will get back to you all a > bit > later... > > Thanks, > K > > > --- Robert Ramey <ramey(a)rrsd.com> wrote: > > For what its worth, changes have been made for 1.33 > which > > work better with > > argument dependent lookup as well as two-phase lookup. > > > > I envisioned that the EXPORT marcros be in the class > > header after the class > > declaration. > > > > For export to work properly, the archive includes have > to > > before the class > > heades with EXPORT in them > > > > All serialization traites macros like EXPORT must be > > invoke from the global > > namespace. > > > > Robert Ramey > > > > > > Jeff Flinn wrote: > > > "kauai" <kauai_diver(a)yahoo.com> wrote in message > > > > > > news:20050510182406.59092.qmail@web60712.mail.yahoo.com... > > >> Hi Robert, > > >> > > >> The EXPORT macros were initially inside a namespace, > > >> interesting that it compiled and linked with gcc. > > >> > > >> After fixing this I still have the same problem with > > >> Microsofts compiler. It complains about the same > > thing as > > >> I state below. Is there some include file that I > may > > be > > >> missing? I can send you some code if that helps. > > > > > > I use: > > > > > > #define BOOST_NO_ARGUMENT_DEPENDENT_LOOKUP > > > > > > previous to including any boost serialization > headers. > > I'm not sure > > > it'll solve your issues, but it did with mine with > > VC7.1. > > > > > > Jeff Flinn > > > > > > > > _______________________________________________ > > Boost-users mailing list > > Boost-users(a)lists.boost.org > > http://lists.boost.org/mailman/listinfo.cgi/boost-users > > > > > > __________________________________ > Yahoo! Mail Mobile > Take Yahoo! Mail with you! Check email on your mobile > phone. > http://mobile.yahoo.com/learn/mail > Yahoo! Mail Stay connected, organized, and protected. Take the tour: http://tour.mail.yahoo.com/mailtour.html

1 0

Question
by Sliwa, Przemyslaw (London) 11 May '05

11 May '05

All, I have a question with regards more to the C++ than Boost. I would like to write a template which depending on the argument type does different things. For example the pseudo code: template <class T, class C> foo(T& nT, C& nC) { // common code; if(T==C) // specific code; ..... // again common code; } Can the template meta programming help me in this case? Sorry, I have no experice with this and I am just guessing. Do you have any suggestions? Thanks for help, Pshemek -------------------------------------------------------- If you are not an intended recipient of this e-mail, please notify the sender, delete it and do not read, act upon, print, disclose, copy, retain or redistribute it. Click here for important additional terms relating to this e-mail. http://www.ml.com/email_terms/ --------------------------------------------------------

2 1

Re: [Boost-users] Re: Re: Re: Re: Boost serializationanddependencynightmares...
by kauai 11 May '05

11 May '05

Hi Robert, Jeff, et all, Thank you Jeff, using that macro helped, I can now build with the Microsoft compiler. Now getting some sort of exception in my serialization code. I am looking into that today, have to setup my visual studio and debugger :) And will get back to you all a bit later... Thanks, K --- Robert Ramey <ramey(a)rrsd.com> wrote: > For what its worth, changes have been made for 1.33 which > work better with > argument dependent lookup as well as two-phase lookup. > > I envisioned that the EXPORT marcros be in the class > header after the class > declaration. > > For export to work properly, the archive includes have to > before the class > heades with EXPORT in them > > All serialization traites macros like EXPORT must be > invoke from the global > namespace. > > Robert Ramey > > > Jeff Flinn wrote: > > "kauai" <kauai_diver(a)yahoo.com> wrote in message > > > news:20050510182406.59092.qmail@web60712.mail.yahoo.com... > >> Hi Robert, > >> > >> The EXPORT macros were initially inside a namespace, > >> interesting that it compiled and linked with gcc. > >> > >> After fixing this I still have the same problem with > >> Microsofts compiler. It complains about the same > thing as > >> I state below. Is there some include file that I may > be > >> missing? I can send you some code if that helps. > > > > I use: > > > > #define BOOST_NO_ARGUMENT_DEPENDENT_LOOKUP > > > > previous to including any boost serialization headers. > I'm not sure > > it'll solve your issues, but it did with mine with > VC7.1. > > > > Jeff Flinn > > > > _______________________________________________ > Boost-users mailing list > Boost-users(a)lists.boost.org > http://lists.boost.org/mailman/listinfo.cgi/boost-users > __________________________________ Yahoo! Mail Mobile Take Yahoo! Mail with you! Check email on your mobile phone. http://mobile.yahoo.com/learn/mail

1 0

Boost Logging Library 1.3 troubles
by Ormond, Randy (Mission Systems) 11 May '05

11 May '05

Hi, If I'm following the threads correctly, it looks like John Torjo's logging library may become an official part of Boost someday. It certainly seems to be perfect for my needs. But I'm having all sorts of trouble on Red Hat with gcc 3.2.2-5. I had to modify some source to avoid compile errors, which has me worried given my lack of expertise. In log.hpp, I had to change (sorry, I can't paste in code or compiler output, this is typed in) template<bool is_compile_time, bool is_enabled> struct logger_keeper_is_enabled... to template<bool is_compile_time, bool is_enabledInput> struct logger_keeper_is_enabled... to avoid shadowing is_enabled in ts_posix.hpp, I had to change the mutex ctr from mutex() : m_mutex(0) {... to mutex() {... This seems OK because m_mutex is initialized in the ctr body. After making those changes, I tried the example in the basic_usage.html. It hangs on the first call to BOOST_DEFINE_LOG. I traced it with gdb to find_log_by_name() at log_manager.cpp:294. From there, it goes to the logger_info ctr. It hangs in lwm_pthreads.hpp:72 in the scoped_lock ctr, apparently trying to get a lock. Of course my mods to the source and general lack of insight have me worried. I expect difficulties as an "early adopter" but any help would be greatly appreciated. BTW, is anyone else trying to use this yet? Thanks, Randy

1 0

RE: [Boost-users] Re: Serialize templates classes > automatique exportdeclaration
by Olivier Vanden Bos 11 May '05

11 May '05

I am not sure of my fact, but can we do it dynamically , as the macro call extended_type_info::self_register() and void extended_type_info::key_register(const char *key_). I ve read in the list some stuff about registering templates like this... template<class> struct pack; template<class T> struct pack<void (T)> { typedef T type; }; BOOST_CLASS_EXPORT( pack<void (DerivedStatus< Algo<int, double > >)>::type ); Can we mix that together to achieve this goal ? In fact, I don't care if we register the type each time we call, I know it s not optimal , but at this point I can live with it... void Add() { B<int>* objectToAdd= new B<int>(0) BoostRegister(B<int>); array.push_back(objectToAdd); } - VDB - "LNK1210: exceeded internal ILK size limit" ________________________________ From: boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org] On Behalf Of Robert Ramey Sent: 11 mai 2005 02:22 To: boost-users(a)lists.boost.org Subject: [Boost-users] Re: Serialize templates classes > automatique exportdeclaration I don't know how to avoid this either. If you figure out how to do this let me know Robert Ramey And when i archive the array in another .cpp the exception is raised, because the class B<int> as not been exported. I would like to avoid explicitly declare BOOST_CLASS_EXPORT_GUID(B<int>, "B_int") For each instanciation of B<> somewhere in the glo bal space.

2 1

FOREACH with maps
by Dave Steffen 11 May '05

11 May '05

Another question about the to-be-released FOREACH macro, which I find I am using heavily... (BTW - Thanks Eric for your quick responses, and for the cool macro!) The issue is declaring the 'loop variable' for maps. For example, vector<int> v; ... FOREACH (int a, v) { ... } // this compiles int b; FOREACH (b, v) {...} // this also compiles However, for maps, the second variation (declaring outside the macro) doesn't work: map<int,int> m; ... FOREACH (map<int,int>::value_type i, m) {...} // compiles map<int,int>::value_type i; FOREACH (i, m) {...} // COMPILE ERROR Specifically, the compiler says error: non-static const member ` const int std::pair<const int,int>::first', can't use default assignment operator (from GCC 3.3.1). Despite my best efforts, I can't figure out what this error means, or what it's talking about. Anybody else run into this? Any thoughts? ---------------------------------------------------------------------- Dave Steffen, Ph.D. "The only justification for our concepts and Software Engineer IV systems of concepts is that they serve to Numerica Corporation represent the complex of our experiences; ph (970) 419-8343 x27 beyond this they have no legitimacy." fax (970) 223-6797 -- Albert Einstein dgsteffen(a)numerica.us

7 9

Unable to create libboost-thread-vc71-mt-sgd-1_31.lib
by Shetty, Diwakar 11 May '05

11 May '05

Hi all, We included the header files of boost in VC++ 7.1 During linking the program automatically tries to link against the static library libboost-thread-vc71-mt-sgd-1_31.lib. However, "bjam install" does not create this library. Though it creates the dll version of the thread library (i.e. thread*.dll and thread*.lib) We tried various combinations such as following to get the static library built. 1) bjam --with-thread -sBUILD="debug <runtime-link>static <threading>single" install 2) bjam --with-thread -sBUILD="debug static single" install 3) bjam -sBUILD="debug static single" install But none of the above worked. We also tried to enable "BOOST_ALL_NO_LIB" macro so that we can specify the dynamic version of the library manually. It did not work. We are using the dynamic version of runtime library. The question I have is, how to build the library "libboost-thread-vc71-mt-sgd-1_31.lib" Any help would be highly appreciated. I have tried every possible thing and seem to be going nowhere with this issue. Thanks, Diwakar

2 1

Fwd: Re: [Boost-users] Re: Boost serialization and dependency nightmares...
by kauai 11 May '05

11 May '05

In addition I resurrected this post to the newsgroup. It is exactly the same problem I am having: boost 1.32, with VC7.1 I'm getting compiler errors in two unrelated translation units both dealing with std::set<std::string> data members. I've gone through and ensured that i've #include'd the <set.hpp> and <string.hpp> headers from boost/serialization. Any ideas on what might be causing this problem? This has worked fine previously with boost 1.31 and serialization20 version of the library. Why would the 3 argument overload not be found? Below are the (beginnings of the ) related compiler errors. Thanks, Jeff ======================== C:\boost\boost_1_32_0\boost\serialization\split_free.hpp(55) : error C2780: 'void boost::archive::load(Archive &,T &)' : expects 2 arguments - 3 provided C:\boost\boost_1_32_0\boost\archive\detail\iserializer.hpp(514) : see declaration of 'boost::archive::load' C:\boost\boost_1_32_0\boost\serialization\split_free.hpp(54) : while compiling class-template member function 'void boost::serialization::free_loader<Archive,T>::invoke(Archive &,T &,const unsigned int)' with [ Archive=boost::archive::text_iarchive, T=std::set<std::string> ] .... ======================== C:\boost\boost_1_32_0\boost\serialization\split_free.hpp(55) : error C2780: 'void boost::archive::load(Archive &,T &)' : expects 2 arguments - 3 provided C:\boost\boost_1_32_0\boost\archive\detail\iserializer.hpp(514) : see declaration of 'boost::archive::load' C:\boost\boost_1_32_0\boost\serialization\split_free.hpp(54) : while compiling class-template member function 'void boost::serialization::free_loader<Archive,T>::invoke(Archive &,T &,const unsigned int)' with [ Archive=boost::archive::binary_iarchive, T=std::set<std::string> ] .... --- kauai <kauai_diver(a)yahoo.com> wrote: > Date: Mon, 9 May 2005 14:07:43 -0700 (PDT) > From: kauai <kauai_diver(a)yahoo.com> > Subject: Re: [Boost-users] Re: Boost serialization and > dependency nightmares... > To: boost-users(a)lists.boost.org > > Robert, > > Thank you. My project requires both linux and Win32 > support. When I use this macro in win32 environment I > get > a compile time error. It generates a fairly long error > which I can send you but here is the first few lines > which > are possibly the most important. > > Again it is fully compiling/linking/running in Linux. > > > cl /nologo /W3 /MDd /Od /EHsc /RTC1 /GS /Gy /Zc:forScope > /GR /Gd /TP /D_DEBUG /D > WIN32 /D_CONSOLE /D_WINDOWS /DUNICODE /DQT_DLL > /DQT_THREAD_SUPPORT > /IW:\vob\abvsw\boost\1_32_0\win32\include\boost-1_32 > /Iroot\abcmn\ /c build-win32-debug\...\xyz.cpp /Fobui > ld-win32-debug\...\xyz.obj xyz.cpp > W:\vob\abvsw\boost\1_32_0\win32\include\boost-1_32\boost\ > serialization\split_free.hpp(55) : error C2780: 'void > boost::archive::load(Archi > ve &,T &)' : expects 2 arguments - 3 provided > W:\vob\abvsw\boost\1_32_0\win32\include\boost-1_3 > 2\boost\archive\detail\iserializer.hpp(514) : see > declaration of 'boost::archive > ::load' > W:\vob\abvsw\boost\1_32_0\win32\include\boost-1_3 > 2\boost\serialization\split_free.hpp(54) : while > compiling > class-template member > function 'void > boost::serialization::free_loader<Archive,T>::invoke(Archive > &,T > &,const unsigned int)' > > When I #ifdef the BOOST_CLASS_EXPORT(ClassName) macros I > compile/link/and run. Of course base class serialzation > is > disable for Win32 at the moment because of this. > > Thanks for any help, > -K > > --- Robert Ramey <ramey(a)rrsd.com> wrote: > > This is addressed with the alternative > BOOST_CLASS_EXPORT > > method. its > > described in the serialization documentation. > > > > Robert Ramey > > > > kauai wrote: > > > Hi all, > > > > > > I have a container class that has a vector which > > contains > > > several other classes. My container class has a > > serialize > > > function that looks like this: > > > > > > friend class boost::serialization::access; > > > template<class Archive> void serialize (Archive &ar, > > const > > > unsigned int version) > > > { > > > // save/load base class information > > > ar & > > BOOST_SERIALIZATION_BASE_OBJECT_NVP(MyBaseClass); > > > ar.register_type(static_cast<SomeClass1*>(NULL)); > > > ar.register_type(static_cast<SomeClass2*>(NULL)); > > > ... > > > ar & BOOST_SERIALIZATION_NVP(m_Vector); // > > std::vector > > > } > > > > > > Got everything working but now an issue has come up: > > > Someone wants to add a new class but it is not part > of > > this > > > package that is doing the serialzation. I want to > > avoid > > > what I call "dependency hell", is there a way for > that > > new > > > class to get registered someplace other than in this > > > function? > > > > > > Thanks for any help, > > > Kauai > > > > > > > > > > > > > > > Discover Yahoo! > > > Stay in touch with email, IM, photo sharing and more. > > Check it out! > > > http://discover.yahoo.com/stayintouch.html > > > > > > > > _______________________________________________ > > Boost-users mailing list > > Boost-users(a)lists.boost.org > > http://lists.boost.org/mailman/listinfo.cgi/boost-users > > > > > > > __________________________________ > Do you Yahoo!? > Yahoo! Mail - You care about security. So do we. > http://promotions.yahoo.com/new_mail > Discover Yahoo! Have fun online with music videos, cool games, IM and more. Check it out! http://discover.yahoo.com/online.html

3 4