- Boost-users - lists.boost.org

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

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