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

Serialize templates classes > automatique export declaration
by Olivier Vanden Bos 11 May '05

11 May '05

I think there have been a lot of discussion about this, It might been related to the template serialization traits in the documentation , but the explanation was a little to subtle for me, \ In fact, I cannot really understand how to avoid The unregistered exception when I use a template like this. // template declaration pseudo code Class A { Virtual ~A(){}; template<class Archive> void serialize(Archive & ar, const unsigned int version){}; }; Template <T> Class B : public A { T Object; B(T& _object) { Object=_object; } template<class Archive> void serialize(Archive & ar, const unsigned int version) { Ar & Object; }; } //Main Code Vector<A*> array; Template<T> void Add(T& val) { B<T>* objectToAdd= new B<T>(val) array.push_back(objectToAdd); } //User code Add((int)0); 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 global space. In the Add() or in the template declaration that s ok, but I cannot suppose what types that the user will use. Thanks in advance.. - VDB - "LNK2019: unresolved external symbol" Olivier Vanden Bos MSN#TheVdeub(a)hotmail.com AI Programmer UBISOFT Montreal

2 1

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

11 May '05

Hi Robert, I've been looking into this further and found some interesting behavior. Don't know if this will help or not, but here goes: * In linux I have my xyz.cpp file, several includes, the archive include, then the export, like this: #include <boost/serialization/export.hpp> // should always be after archive includes After that I have: BOOST_CLASS_EXPORT(Foo::Bar::MyClass); using namespace Foo::Bar; // the rest of the file. This code plus the rest will compile, link, and run properly in Linux using gcc 3.2.2. (I can get more version info if you need, kernel, glibc, etc.) * In Windows Server 20003 with cl version 13.10.3077 I can not compile. I get the same error message I stated earlier. Now the wierd part: If I remove the #include <boost/serialization/export.hpp> I cannot compile anymore with gcc, as expected. There error I get is this: ISO C++ forbids declaration of `BOOST_CLASS_EXPORT' with no type In Windows not only does it compile, but it links and runs. 'runs' meaning there is an .exe file but my serialization test code still fails ;) I dont know if these annoying details may help but I just thought I'd share. -K --- kauai <kauai_diver(a)yahoo.com> wrote: > Date: Tue, 10 May 2005 11:24:06 -0700 (PDT) > From: kauai <kauai_diver(a)yahoo.com> > Subject: Re: [Boost-users] Re: Re: Re: Boost > serialization and dependencynightmares... > To: boost-users(a)lists.boost.org > > 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. > > Also where is the best place to call these macros? At > the > moment they are at the top in the appropriate .cpp files. > > Thanks, > -K > > > --- Robert Ramey <ramey(a)rrsd.com> wrote: > > > Just a guess, but are the EXPORT macros being used in > > side a namespace? > > They should be outside of any namespace. > > > > Robert Ramey > > > > kauai wrote: > > > 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 > > >>> > > >>> > > >>> > > >>> _______________________________________________ > > > === message truncated === > > > __________________________________________________ > Do You Yahoo!? > Tired of spam? Yahoo! Mail has the best spam protection > around > http://mail.yahoo.com > Discover Yahoo! Find restaurants, movies, travel and more fun for the weekend. Check it out! http://discover.yahoo.com/weekend.html

2 1

creating static functions with metaprogramming
by Felipe Magno de Almeida 11 May '05

11 May '05

I would like to know if it is possible to generate a number of static functions that redirect to some other function inside the class, taking an object as void * and is casted to the object (that it knows which is), and be possible to take the address of such functions... well, i dont have a good english, so I will reexplain: For the ones that already are familiar with Win32 GUI: I must create a static function to be the DialogProc of a window, the problem is... I'm creating some property sheets to be used with one class, and I would like to make it automatic the redirection from the static function to some other method inside the class, that isnt static... and I would like to create a class, probably with CRTP to say to it to create the functions that would redirect to me, saying how much functions would that be, and make it creates the codes that gives to windows the static functions addresses and create the functions that would redirect.... Then I just needed to create my functions that they redirect to me... making it a LOT easier to me, and would make my code a lot more concise... Sorry for the off-topic... But I think you would be the best ones to direct this question.... Thanks for the attention, Felipe Magno de Almeida -- Felipe Magno de Almeida UIN: 2113442 email: felipe.m.almeida+spam at gmail com felipe.almeida at ic unicamp br, felipe.m.almeida at gmail com, felipe at synergy com I am a C, modern C++, MFC, ODBC, Windows Services, MAPI developer from synergy, and Computer Science student from State University of Campinas(UNICAMP). To know more about: Unicamp: http://www.ic.unicamp.br Synergy: http://www.synergy.com.br current work: http://www.mintercept.com "There is no dark side of the moon really. Matter of fact it's all dark."

2 2

regex lib - using replace
by Lucian Ion 10 May '05

10 May '05

Hi, I need one regular expression and one format string to do the following 1) if text starts with s1, replace s1 with r1 2) else, if text starts with s2, replace s2 with r2 3) else, if text starts with s3, replace s2 with r3 4) else, prepend r3 to text While I managed to work out the first three using an expression like (^s1)|(^s2)|(^s3) and the format as 1?r1:2?r2:3?r3: I have no idea how to include the fourth case into my expression. Any suggestion? Thank you in advance, Lucian

2 1

Re: regex lib - using replace
by Lucian Ion 10 May '05

10 May '05

sorry, I mean 4) else, prepend r4 to text (that is, has nothing to do with the other replacement strings) ----- Original Message ----- From: "Lucian Ion" <lucianion(a)rapidsolution.ro> To: <boost-users(a)lists.boost.org> Sent: Tuesday, May 10, 2005 8:50 PM Subject: regex lib - using replace > Hi, > I need one regular expression and one format string to do the following > 1) if text starts with s1, replace s1 with r1 > 2) else, if text starts with s2, replace s2 with r2 > 3) else, if text starts with s3, replace s2 with r3 > 4) else, prepend r3 to text > > While I managed to work out the first three using an expression like > (^s1)|(^s2)|(^s3) and the format as 1?r1:2?r2:3?r3: > I have no idea how to include the fourth case into my expression. Any > suggestion? > > Thank you in advance, > Lucian >

1 0

Boos.Bind: Binding to a placeholder address
by Adi Shavit 10 May '05

10 May '05

Hi, Say have function I want to bind that expects an int*, but my placeholder _1 is of type int. How can I tell the bind function to bind "&_1"? I guess it would be something like: bind(myFun, address_of(_1)); Is this even possible? Thanks, Adi

3 3

Standard Library Extension TR2 Call for Proposals
by Beman Dawes 10 May '05

10 May '05

The C++ Standards Committee has begun work on a second library technical report. A Call for Proposals is available at http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2005/n1810.html --Beman

1 0

Building Boost on MacOS 10.4 (Tiger)
by Dave Snowdon 10 May '05

10 May '05

Hello Has anyone successfully built boost on an Mac running MacOS 10.4? If so I'd appreciate some pointers on what worked for you. I've tried both boost 1.32 and a "fresh from CVS" copy with gcc 3.3 and gcc 4.0 on a Powerbook G4 with 512Mb RAM with a fresh install of MacOS 10.4. The Mac has 1.15Gb free on the startup partition and 1.36Gb free on the partition where I'm compiling boost. So far no attempt has met with success. All builds were attempted from the Terminal app with no other applications runnng. Compiling boost 1.32 with gcc 4.0 (the default compiler on MacOS 10.4) fails (unrecognised compiler version) because at the time 1.32 was released gcc 4.0 was not available. Compiling boost 1.32 with gcc 3.3 (bjam "-sTOOLS=darwin" "-sGCC=gcc-3.3" "-sGXX=gcc-3.3" stage) gives some rather strange results. Link actions seems to fail such as: ...failed darwin-Link-DyLib-action bin/boost/libs/date_time/build/libboost_date_time.dylib/darwin/debug/shared-linkable-true/libboost_date_time-d-1_32.dylib... Almost all the calls to ld gives undefined symbol errors similar to (I'm not sure if this is due to previously failed link actions): darwin-Link-DyLib-action bin/boost/libs/date_time/build/libboost_date_time.dylib/darwin/debug/shared-linkable-true/libboost_date_time-d-1_32.dylib ld: Undefined symbols: __ZNKSbIwSt11char_traitsIwESaIwEE3endEv __ZNKSbIwSt11char_traitsIwESaIwEE5beginEv __ZNKSs2atEm __ZNKSs3endEv __ZNKSs5beginEv __ZNKSs6lengthEv __ZNKSs7compareERKSs __ZNKSt11logic_error4whatEv __ZNKSt18basic_stringstreamIcSt11char_traitsIcESaIcEE3strEv __ZNKSt18basic_stringstreamIwSt11char_traitsIwESaIwEE3strEv __ZNKSt9basic_iosIcSt11char_traitsIcEE4failEv __ZNKSt9basic_iosIwSt11char_traitsIwEE4failEv __ZNKSt9type_infoeqERKS_ __ZNSaIcEC1Ev __ZNSaIcED1Ev After about 15 minutes or so the Mac grinds to a halt and starts popping up dialog boxes warning that the startup disk is full and then later that "Your Mac OS X startup disk has no more space available for application memory. To avoid problems with your computer quit any applications you may not be using" Using gcc 4.0 (bjam "-sTOOLS=darwin" stage) on the the CVS copy gives many errors similar to this: ...failed darwin-C++-action bin/boost/libs/wave/build/libboost_wave.a/darwin/release/cpp_re.o... ...skipped <@boost!libs!wave!build/libboost_wave.a/darwin/release>libboost_wave-1_33.a for lack of <@boost!libs!wave!build/libboost_wave.a/darwin/release>instantiate_cpp_exprgrammar.o... ...skipped <@boost!lib>libboost_wave-1_33.a for lack of <@boost!libs!wave!build/libboost_wave.a/darwin/release>libboost_wave-1_33.a... Previously I've been using boost 1.32 on a machine running Red Hat Linux 8 with gcc 3.2 and have not experienced any problems getting it to build. Any advice appreciated.... thanks Dave

2 1

Need Help
by pankaj jain 10 May '05

10 May '05

Hi , I am new user of boost.python. Please help me in solving this problem. problem 1) When I am running bjam from home/boost/libs/python/example/tutorial folder its running fine but building my hello.so in home/boost/bin/boost/libs/python/example/tutorial/hello.so/gcc/.. folder.is there any way to chagne the path for hello.so problem 2): when I am running bjam from any other folder ( home/boost/sample) its giving me this error :: /home/jamrules no such directory .../.../allyourbase.jam 1451 see definition of rules "root-path: being called. Please let me know any solution. Thanks for all your help Pankaj Joaqu�n M� L�pez Mu�oz <joaquin(a)tid.es> wrote: Dave Snowdon ha escrito: > I'd like to second what Stefan said. Before I started using boost (only > a couple of months back) I'd seen references to it in many articles in > the C++ User Journal not to mention a few C++ related books. Maybe that's a significant difference wrt to the rest of the C++ community: Many programmers are not subscribed to CUJ and do not buy intermediate/advanced books (entry-level references do not usually list Boost.) I'd like a website that holds the attention of the casual comer. A newbie with just a passing knowledge of what Boost is might be scared by the contents of boost.org and never return back; a more friendly site, with forums and stuff, is a more agreeable place to hang around and eventually decide to try Boost for real. Joaqu�n M L�pez Mu�oz Telef�nica, Investigaci�n y Desarrollo > After > starting on a new C++ project I realised I needed some of the libraries > mentioned and went straight for them. So it was not so much a case of > "do I need boost?" but "where do I find the bits I need now, and how do > I use them?" > > Dave > > Stefan Strasser wrote: > > > I don't think people come by accident to the boost page(and they > > definitely don't to the community page) and need to be convinced that > > they need boost. > > they look for a specific library/class and they've been told, "boost > > has that". later on they look what also is in boost. > > at least that is how it worked for me. > > > > so a community page(especially user comments in documentation) helps a > > new user to get into it. it is imho not the function of a community > > page to raise interest. (and I think it can't). > > > > and I think the best thing you can do to bring a new user into the > > community thing is to have documentation and user comments on the same > > page(see php). because what a user definitely does is look at the > > documentation when he's found the library he's looking for. > > > > _______________________________________________ > Boost-users mailing list > Boost-users(a)lists.boost.org > http://lists.boost.org/mailman/listinfo.cgi/boost-users _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users --------------------------------- Yahoo! Mail Stay connected, organized, and protected. Take the tour

2 1

order of pointer serialisation
by martin f krafft 10 May '05

10 May '05

Consider a doubly-linked list. How do you serialise this? The natural way would be to provide a serialize function such as: template <class Archive> void serialize(Archive& ar, unsigned int const) { ar & prev; ar & next; } This, however, causes "pointer conflict" exception to be thrown, because the next pointers are serialised before the actual list node is serialised. The following is an abridged scneario illustrating the same dilemma: class Foo { friend class boost::serialization::access; template <typename AT> void serialize(AT& ar, unsigned int const) {} }; int main(void) { Foo x; Foo *y = &x; boost::archive::text_oarchive oa(std::cout); oa & y; // serialise the pointer before the actual instance oa & x; // throws boost::archive::archive_exception(pointer_conflict) return 0; } I could come up with many more examples in which it is now granted that an instance has been serialised by the time that a pointer to that very instance needs to be serialised. How does boost::serialization expect this challenge to be handled? Note that I understand very well what pointer_conflict is all about. I do wonder whether this protection is always necessary. Worse yet: there seems to be no way to disable it; catching the pointer_conflict skips over the serialisation of the instance. Is the only way to deal with this to ensure that instances are always saved first? -- martin; (greetings from the heart of the sun.) \____ echo mailto: !#^."<*>"|tr "<*> mailto:" net@madduck invalid/expired pgp subkeys? use subkeys.pgp.net as keyserver! spamtraps: madduck.bogus(a)madduck.net micro$oft encrypts your windows nt password when stored on a windows ce device. but if you look carefully at their encryption algorithm, they simply xor the password with "susageP", Pegasus spelled backwards. Pegasus is the code name of windows ce. this is so pathetic it's staggering.

1 0