- Boost-users - lists.boost.org

Re: [Boost-users] Re: (de)serializing vector<const A*>
by Zoltan Szatmary 19 Jan '05

19 Jan '05

Dear Robert Thank you for answering. I am sorry I was not very specific (though please don't think I would put the question without RTFM first ;) ) The part of documentation you pointed to told me to do const_cast manually. It is fine in case of const members of a class, but this time the const type is contained by a vector. I could manually write the serialization code for my vector<const A*> type but I hoped I could use the built-in vector serialization templates in vector.hpp To be clear I copy here the piece of code I am talking about #include <fstream> #include <boost/archive/text_iarchive.hpp> #include <boost/archive/text_oarchive.hpp> #include <boost/serialization/vector.hpp> using namespace std; class A { friend class boost::serialization::access; template<class Archive> void serialize (Archive &ar, const unsigned int version) { } }; /* NEITHER OF THESE HELPS namespace boost { namespace serialization { template <class Archive> inline void serialize (Archive &ar, const A *p, const unsigned int version) { ar & const_cast<A*>(p); } template <class Archive> inline void serialize (Archive &ar, const A &p, const unsigned int version) { ar & const_cast<A&>(p); } }} */ int main() { A a; vector<const A*> v(1, &a); ofstream ofs ("vectorofconstptr"); boost::archive::text_oarchive oa(ofs); oa << v; ofs.close(); // UNTIL THIS POINT EVERYTHING IS FINE // BUT DESERIALIZATION FAILS TO COMPILE ifstream ifs("vectorofconstptr"); boost::archive::text_iarchive ia(ifs); vector<const A*> p; ia >> p; ifs.close(); } Please tell me if I am doing something wrong. I tried to look into the documentation, I have even followed the serialization library code for a while, but eventually decided to ask. Thank you in advance Zoltan >Look in the documentation under Reference/Class Serialization/const members >Robert Ramey >Zoltan Szatmary wrote: >> Hello >> >> could you help please? what is the easiest way to (de)serialize a >> member of type vector<const A*> (A being a base class)? Serializing >> is fine, but deserialization fails to compile since the vector >> deserialization code tries to deserialize a const A* pointer without >> const_cast-ing it (I guess). >> >> Thank you in advance >> >> Zoltan

2 1

long + bigint
by Benjamin Schmeling 19 Jan '05

19 Jan '05

Hi, I am wrapping a class "bigint" from c++ to python. I have overloaded the add method to do something like this on python: >>>n=bigint(5); >>>n+5L 5 This works well, but if I try to do it vice versa my overloaded add method isn't called by python: >>>n=bigint(5); >>>5L+n Causes error. My add methods look like this: bigint add_bigint_l(bigint const& b, long_ const& l) bigint add_l_bigint(long_ const& l,bigint const& b) and are defined by .def("__add__", add_bigint_l) .def("__add__", add_l_bigint) Can someone tell me howto realize it? Benjamin Schmeling -- GMX im TV ... Die Gedanken sind frei ... Schon gesehen? Jetzt Spot online ansehen: http://www.gmx.net/de/go/tv-spot

2 1

re: strange filesystem dll behaviour
by Paul 19 Jan '05

19 Jan '05

Bugger, I forgot to mention: I'm using boost 1.32 and GCC 3.4.2 on a Windows 2000 computer. Thanks Paul Find local movie times and trailers on Yahoo! Movies. http://au.movies.yahoo.com

1 0

RE: [Boost-users] boost::bind and methods of managed classes
by Bjorn.Karlsson＠readsoft.com 19 Jan '05

19 Jan '05

Hello Vasco, > From: Vasco Lohrenscheit [mailto:valoh@krautgames.de] > > is there a easy way/workaround to use methods of managed classes with > boost::bind? It's not exactly easy, but I think it can be done. First of all, you won't be able to take the address of TestManagedClass::test because it's a managed class. Second, the calling conventions of invoking the signal (default: __stdcall) and the function test (__clrcall) are different. Here's a workaround that should work: * Create a nested __nogc class in the __gc class (technically, it doesn't need to be nested, just __nogc). * In the nested class, store a pointer (using gcroot<>, as it's a managed class) to the outer class. * In the constructor of the outer class, create an instance of the nested class. * Define a callback member in both classes. In the inner class, simply forward to the corresponding function in the outer class. * Create the bind expression using the address of the nested class' function. * Invoke the signal and keep your fingers crossed. Here's an example that demonstrates the technique: #include "boost/bind.hpp" #include "boost/signals.hpp" #include "vcclr.h" // For gcroot<> boost::signal<void (int)> sig; __gc class managed_class { public: managed_class() : inner_(new unmanaged_class(this)) { // Connect to the signal sig.connect(boost::bind(&managed_class::unmanaged_class::callback,inner_,_1) ); } void callback(int i) { // This is the managed "callback" } private: // Here's the nested class __nogc class unmanaged_class { public: unmanaged_class(managed_class* outer) : outer_(outer) {} void callback(int i) { outer_->callback(i); } private: gcroot<managed_class*> outer_; }; unmanaged_class* inner_; }; void main() { managed_class* p=new managed_class(); sig(42); // Invokes the signal, // which invokes the binder, // which invokes unmanaged_class::callback(), // which invokes managed_class::callback(). } Bjorn Karlsson

1 0

[Boost-users] Compile on OSX hangs, huge memory usage, xml_grammar.o
by Stian Soiland 19 Jan '05

19 Jan '05

Tried to build boost_1_32_0 on OSX today, but the build process brought the machine to trashing, as cc1plus ended up using all available memory. The problem seems related to libs/serialization - removing it made the rest of boost build almost OK. Here are the details: Mac OS X 10.3.7, with gcc version 3.3 20030304 (Apple Computer, Inc. build 1666) Lots of programs from fink installed to /sw Lots of programs from opendarwin installed to /opt 1. port install boost-jam (3.1.10) 2. Downloaded boost_1_32_0.tar.bz2 from boost.org and unpacked 3. bjam -sTOOLS=darwin stage 4. Start top in another terminal 5. Wait some hours until your hard drive starts to make crazy noises 6. Last bjam output will be: darwin-C++-action bin/boost/libs/serialization/build/libboost_serialization.a/darwin/ release/xml_grammar.o Top will show something like: Processes: 88 total, 2 running, 1 stuck, 85 sleeping... 223 threads 17:46:32 Load Avg: 1.72, 1.89, 1.45 CPU usage: 9.5% user, 30.2% sys, 60.3% idle SharedLibs: num = 131, resident = 7.25M code, 404K data, 136K LinkEdit MemRegions: num = 28961, resident = 896M + 2.28M private, 16.7M shared PhysMem: 111M wired, 601M active, 300M inactive, 1013M used, 10.6M free VM: 7.65G + 88.6M 1116924(79) pageins, 1865004(459) pageouts PID COMMAND %CPU TIME #TH #PRTS #MREGS RPRVT RSHRD RSIZE VSIZE 28631 cc1plus 11.5% 1:51.55 1 12 16184 842M- 804K 543M- 1.09G+ 28630 g++-3.3 0.0% 0:00.03 1 11 18 0K 264K 76K 21.8M 28628 bjam 0.0% 0:07.44 1 12 52 4K 264K 92K 31.2M rm -rf libs/serialization to avoid problem. But I guess some people actually need those xml-grammars.. -- Stian Søiland You can't say civilization don't Trondheim, Norway advance, however, for in every war http://www.soiland.no/ they kill you in a new way. [Rogers] =/\=

2 1

[stringalgo] Tokenizer with string delimiters?
by Stefan Slapeta 19 Jan '05

19 Jan '05

Hi, please could anybody tell if it's possible to use split() to tokenize a string where the token delimiters are strings of - n characters, e.g. "\r\n" - m..n characters, e.g. "\n" _or_ "\r\n" If not, are there alternatives? Thanks, Stefan

2 1

Re: [Boost-users] Re: Re: gregoran::year_based_generator and=?ISO-8859-1?Q? mingw-3.4.2?=
by Antonio Coralles 19 Jan '05

19 Jan '05

boost-users(a)lists.boost.org wrote: > On Tue, 18 Jan 2005 17:19:56 -0700, Bart wrote > >> > i've also tried to compile some other examples - they work, like the >> > rest of my program; maybe there is something wrong with my > installation >> > - but i don't know what. it is also really strange that mingw accepts >> > partial_date [in my app] which is derived from year_based_generator > but >> > refuses to work with year_based_generator*. >> > >> > antonio >> >> I'm stumped. >> >> I tried upgrading to gcc/g++ 3.4 (I extracted these into my original >> mingw installation) and am unable to duplicate your errors. >> >> If you can spare the time, try downloading a fresh boost_1_32_0 archive >> and attempting a print_holidays compile with that source tree. I'm >> wondering if your source tree may have been corrupted somehow. > > That would be my guess at this point too. The fact that the example fails > means something is really off. I just compiled 1_32 with mingw dev-cpp > with > compiler version 3.3.1. The regression tests also succeed for your > combination. More than likely it would have to be something messed up in > boost/date_time/date_generators.hpp -- so you could try just getting the > latest version of that file and replacing it. > > > Jeff i'll try a complete reinstallation of the boost-libraries in the next few days. i think it is more likely that the library isn't correctly installed because i had the same problems with 1_31_0. it is eaven possible that there is something wrong with my mingw installation. antonio

1 0

(de)serializing vector<const A*>
by Zoltan Szatmary 19 Jan '05

19 Jan '05

Hello could you help please? what is the easiest way to (de)serialize a member of type vector<const A*> (A being a base class)? Serializing is fine, but deserialization fails to compile since the vector deserialization code tries to deserialize a const A* pointer without const_cast-ing it (I guess). Thank you in advance Zoltan

2 1

Re: [Boost-users] Re: gregoran::year_based_generator and mingw-3.4.2
by Antonio Coralles 19 Jan '05

19 Jan '05

[snip] > boost-users(a)lists.boost.org <mailto:boost-users@lists.boost.org> wrote: >> "Antonio Coralles" <noche.suapie(a)reflex.at > <mailto:noche.suapie@reflex.at> <mailto:>> wrote in message >> news:41EBE2F0.7020003@reflex.at... >>> i'm writing on a programm which uses the boost date_time library >>> [boost-version = 1.32.0] heavily; while i normally programm in linux >> using >>> gcc-3.3.5 i also need to compile my program in windows from time to >> time. >>> because it makes life easier with another library [gtkmm-2.4.8] i'm >> using >>> mingw-3.4.2; everthing worked fine (my code contains about 10 000 lines >>> using boost_date_time) until i started using a >>> boost::gregorian::year_based_generator. while gcc-3.3.5 in liux >> accepts my >>> code without even a single warning [with -Wall], mingw-3.4.2 complains - >>> it seems that the compiler has problems with year_based_generator: > [snip] [snip] >> Are you able to build the print_holidays example? It uses pointers to >> year_base_generators. [snip] > > i'll try that in the next time and post the results of this experiment; > be the way, i've now compiled my app in win succesfully by putting > evrything which is somehow connected to year_based_generator into > #ifndef WIN32_COMPILE .... #endif. that isn't too bad for the moment, > because i don't use these methods by now in > other parts of my prog [except unit tests]. but sooner or later i will ... > [snip] i've now tried to compile the print_holidays example: this is what i got: $ g++ -lboost_date_time-mgw-1_32 print_holidays.cpp -o test.exe print_holidays.cpp: In function `int main()': print_holidays.cpp:43: error: `year_based_generator' undeclared (first use this function) print_holidays.cpp:43: error: (Each undeclared identifier is reported only once for each function it appears in.) print_holidays.cpp:43: error: template argument 1 is invalid print_holidays.cpp:43: error: template argument 2 is invalid print_holidays.cpp:43: error: invalid type in declaration before ';' token print_holidays.cpp:44: error: `push_back' has not been declared print_holidays.cpp:44: error: request for member of non-aggregate type before '( ' token print_holidays.cpp:45: error: `push_back' has not been declared print_holidays.cpp:45: error: request for member of non-aggregate type before '( ' token print_holidays.cpp:46: error: `push_back' has not been declared print_holidays.cpp:46: error: request for member of non-aggregate type before '( ' token print_holidays.cpp:53: error: `push_back' has not been declared print_holidays.cpp:53: error: request for member of non-aggregate type before '( ' token print_holidays.cpp:55: error: `push_back' has not been declared print_holidays.cpp:55: error: request for member of non-aggregate type before '( ' token print_holidays.cpp:57: error: `push_back' has not been declared print_holidays.cpp:57: error: request for member of non-aggregate type before '( ' token print_holidays.cpp:59: error: `push_back' has not been declared print_holidays.cpp:59: error: request for member of non-aggregate type before '( ' token print_holidays.cpp:64: error: template argument 1 is invalid print_holidays.cpp:64: error: template argument 2 is invalid print_holidays.cpp:64: error: expected `;' before '::' token print_holidays.cpp:65: error: `it' undeclared (first use this function) print_holidays.cpp:65: error: `end' has not been declared print_holidays.cpp:65: error: request for member of non-aggregate type before '( ' token ' token i've also tried to compile some other examples - they work, like the rest of my program; maybe there is something wrong with my installation - but i don't know what. it is also really strange that mingw accepts partial_date [in my app] which is derived from year_based_generator but refuses to work with year_based_generator*. antonio

3 2

mpl::sequence question
by Martin Pasdzierny 19 Jan '05

19 Jan '05

Hello meta programmers, I am looking for a way to extract a sub sequence of a given type list by identifying the begin/end of the sub sequence by the corresponding iterators <code> #include <boost/mpl/vector.hpp> #include <boost/mpl/for_each.hpp> #include <boost/mpl/less.hpp> #include <boost/mpl/greater.hpp> #include <boost/mpl/and.hpp> #include <boost/mpl/find.hpp> #include <boost/mpl/lambda_fwd.hpp> #include <boost/mpl/lambda.hpp> #include <boost/mpl/placeholders.hpp> #include <boost/mpl/if.hpp> #include <boost/mpl/copy.hpp> #include <boost/mpl/copy_if.hpp> #include <boost/mpl/remove_if.hpp> #include <boost/mpl/back_inserter.hpp> #include <boost/mpl/transform.hpp> #include <boost/mpl/push_back.hpp> #include <boost/mpl/int.hpp> // these namespace clauses are motivated from "C++ Template Metaprogramming" by D.Abrahams,A.Gurtovoy: namespace mpl = boost::mpl ; using mpl::placeholders; // But why do I still need to type 'mpl::_1' instead of plain '_1' ? // here is my type list typedef mpl::vector<int,double,std::string > my_types; // from_pos and to_pos are start and end iterators of the subsequence typedef typename mpl::find<my_types,int>::type from_pos; typedef typename mpl::find<my_types,std::string>::type to_pos; // all elements (types) of my_types should be copied into the new seuqence my_sub_types if their iterator (index) is inside the given interval (from_pos,to_pos) typedef mpl::copy_if<my_types, mpl::and_< mpl::greater< mpl::find<my_types,mpl::_1>::type, mpl::int_<from_pos> > , mpl::less< mpl::find<my_types,mpl::_1>::type, mpl::int_<to_pos> > >, mpl::back_inserter< mpl::vector<> > >::type my_sub_types; </code> This code (and lots of slight variations) does not compile. Why ? Any hint (conceptual,syntactical,etc.) is welcome ! regards,Martin. <martin(a)pasdzierny.de>

2 1