September 2008 - Boost-users

[statechart] split different orthogonal regions over different translation units
by Manuel Jung 11 Sep '08

11 Sep '08

Hey, Im trying to split the orthogonal regions over differnent translation units, but it doesn't compile. The compiler complains that the type is incomplete, which i try to put on a different translation unit. What i do is: Forward declare the orthogonal region and define than the outermost state "Active" like this: struct StateInDifferentTransUnit; struct Active; struct Start : sc::asynchronous_state_machine< Start, Active >{ ... } struct Active : sc::state< Active, Start, mpl::list< State1, State2, StateInDifferentTransUnit > >{ ... } Is the split in different translationunits not possible with orthogonal regions or what could be my problem? Kind Regards Manuel

2 1

[thread] working directory specific to each thread
by frederic.bron＠alcan.com 11 Sep '08

11 Sep '08

Is there a way to have the working directory specific to each thread? I have tried to change the working directory in each thread but the working directory seems to be the same in each thread, the last to be set. F. Bron Avis : Ce message et toute pièce jointe sont la propriété d'Alcan et sont destinés seulement aux personnes ou à l'entité à qui le message est adressé. Si vous avez reçu ce message par erreur, veuillez le détruire et en aviser l'expéditeur par courriel. Si vous n'êtes pas le destinataire du message, vous n'êtes pas autorisé à utiliser, à copier ou à divulguer le contenu du message ou ses pièces jointes en tout ou en partie. Notice: This message and any attachments are the property of Alcan and are intended solely for the named recipients or entity to whom this message is addressed. If you have received this message in error please inform the sender via e-mail and destroy the message. If you are not the intended recipient you are not allowed to use, copy or disclose the contents or attachments in whole or in part.

2 1

Re: [Boost-users] [bind] [multi-index] Create type of pointer to overloaded erase method
by JOAQUIN M. LOPEZ MUÑOZ 11 Sep '08

11 Sep '08

________________________________________ De: boost-users-bounces(a)lists.boost.org [boost-users-bounces(a)lists.boost.org] En nombre de JOAQUIN M. LOPEZ MUÑOZ [joaquin(a)tid.es] Enviado el: miércoles, 10 de septiembre de 2008 20:34 Para: boost-users(a)lists.boost.org Asunto: Re: [Boost-users] [bind] [multi-index] Create type of pointer to overloaded erase method > > ________________________________________ > De: boost-users-bounces(a)lists.boost.org [boost-users-bounces(a)lists.boost.org] En nombre de Steven > Watanabe [watanabesj(a)gmail.com] > Enviado el: miércoles, 10 de septiembre de 2008 18:54 > Para: boost-users(a)lists.boost.org > Asunto: Re: [Boost-users] [bind] [multi-index] Create type of pointer to overloaded erase method > > > AMDG > > > > Peter Barker wrote: > > > RecordsContainer::size_type (RecordsContainer::*erase)(unsigned int) = > > > &RecordsContainer::erase; > > > > > > but it failed to compile. Visual Studio 2003 gives me the following > > > error on the above line: > > > <snip> > > > > > > > The following compiles for me with msvc 9.0 > > > > int main() { > > RecordsContainer::size_type (RecordsContainer::*erase)(unsigned int) = > > &RecordsContainer::erase; > > } > > Oddly enough, it fails with MSVC++ 8.0 (though it shouldn't since your > call_traits analysis is correct): > > error C2440: 'initializing' : cannot convert from 'overloaded-function' to > 'boost::multi_index::detail::ordered_index<KeyFromValue,Compare,SuperMeta, > TagList,Category>::size_type (__thiscall > boost::multi_index::multi_index_container<Value,IndexSpecifierList>::* )(unsigned int)' But the following works: int main() { typedef RecordsContainer::index<RecsByID>::type RecordsContainerByID; RecordsContainerByID::size_type (RecordsContainerByID::*erase)(unsigned int) = &RecordsContainerByID::erase; } even though my previous approach is IMHO still better. Joaquín M López Muñoz Telefónica, Investigación y Desarrollo

3 3

[bind] [multi-index] Create type of pointer to overloaded erase method
by Peter Barker 11 Sep '08

11 Sep '08

Hello all, I have a container of the following type: struct RecsByID {}; struct RecsBySizeAndFit {}; typedef boost::multi_index_container< boost::shared_ptr<LastSizeFitRecord>, bmi::indexed_by< bmi::ordered_unique<bmi::tag<RecsByID>,bmi::const_mem_fun<RecordBase,unsigned int,&RecordBase::getRecID> >, bmi::ordered_unique<bmi::tag<RecsBySizeAndFit>,bmi::const_mem_fun<LastSizeFitRecord,const SizeAndFit&,&LastSizeFitRecord::getSizeAndFit> > > > RecordsContainer; The main thing to note above is that the first index is of type "unsigned int" - the record ID. I have a container of unsigned ints which represent the records to be deleted so naturally wanted to do a std::for_each() with a binder and passing in _1 to RecordsContainer::erase. Not that simple because Records::Container::erase has three overloads so I'm trying to come up with a type that will represent the overload I want. I've tried this: RecordsContainer::size_type (RecordsContainer::*erase)(unsigned int) = &RecordsContainer::erase; but it failed to compile. Visual Studio 2003 gives me the following error on the above line: error C2440: 'initializing' : cannot convert from 'overloaded-function' to '<Unknown>' I've decided this particular pointer to member would not be intuitive to use in production code, but I thought I'd try to get it working and failed and would like to know what I'm doing wrong! Regards, Pete

2 2

boost parameter - overloading on keyword "signature"
by e r 10 Sep '08

10 Sep '08

hello, i have something like this: *hpp : namespace tag{ template<typename Id> struct sequence{ static ::boost::parameter::keyword< a_<Id> >& a; static ::boost::parameter::keyword< b_<Id> >& b; static ::boost::parameter::keyword< arg_method_<Id> >& arg_method; }; } namespace sequence_arg{ struct method{}; struct a:public method{}; struct b:public method{}; } class sequence{ public: template<typename Id,typename ArgPack> sequence(const Id& id,const ArgPack&args){ fill_elements(id,args,args[tag::sequence<Id>::arg_method]); } private: template<typename Id,typename ArgPack> void fill_elements( const Id& id, const ArgPack& args, sequence_arg::a method ); template<typename Id,typename ArgPack> void fill_elements( const Id& id, const ArgPack& args, sequence_arg::b method ); }; *.cpp : sequence seq( id(), ( tag::sequence::a = a_value, tag::sequence::arg_method = sequence_arg::a() ) ); So far so good except there is redundancy of information about which method to use, so I'm looking for a way to be able to call say sequence seq( id(), ( tag::sequence::a = a_value ) ) Any suggestion?

2 2

boost fusion and/or boost parameter: open ended question about constructing object from a string
by e r 10 Sep '08

10 Sep '08

hi, here's what i'd like to do A] hpp: struct my{ template<typename ArgPack> my(const ArgPack& args):n_(args[tag::n]),k_(args[tag::k]){} } struct argpack{ arglist<???> operator()(const string& str){???} } cpp: string str("n=2; k=1"); my(argpack(str)); In words,I'm looking to construct an ArgPack from a string, what's the appropriate library for this? Does it seem like I should invest in Boost Spirit? B] The above is a simplified version of what i really need, which is, potentially, 1) having to construct a large number of objects (to be pushed at the back of a container) 2) each object taking a large number arguments (not just n and k as above) 3) each object an instance of one of a set of classes (my0,my1,my2,...,myN) each with its specific argument tags (keywords). So I'd like to put all these arguments in a file, each line corresponding to an object, rather than have a messy cpp file. this is all very open-ended, but so reads the subject title... thanks!

2 5

fusion/adapted/mpl.hpp
by e r 10 Sep '08

10 Sep '08

Hi, I would need this: // a) typedef fusion::vector<...> input0_type; typedef mpl::transform( input0_type, F0 )::type result0_type; //b) typedef fusion::vector<...> resul1_type; result1_type result1 = fusion::transform( result0_type(), f1 ); Of course a) is not allowed, but I guess I can do: // a) typedef mpl::vector<...> mpl_input0_type; //same elements as input0_type typedef mpl::transform( mpl_input0_type, F0 )::type mpl_result0_type; //b) #include <boost/fusion/adapted/mpl.hpp> #include <boost/fusion/include/mpl.hpp> typedef fusion::vector<...> resul1_type; result1_type result1 = fusion::transform( mpl_result0_type(),//? f1 ); in my case, result1_type = input0_type. So I have to define input0_type and mpl_input0_type separately although they convey the same information. i'd like to get rid of this redundancy. I wished there existed a metafun fusion_to_mpl such that MPL_ASSERT((is_same<fusion_to_mpl<input0_type>::type,mpl_input0_type>)) so that the 1st version of a) could be replaced by typedef mpl::transform( fusion_to_mpl<input0_type>::type, F0 )::type mpl_result0_type; and keep the second version of b) any help appreciated. thanks!

2 2

Re: [Boost-users] [bind] [multi-index] Create type of pointer to overloaded erase method
by JOAQUIN M. LOPEZ MUÑOZ 10 Sep '08

10 Sep '08

________________________________________ De: boost-users-bounces(a)lists.boost.org [boost-users-bounces(a)lists.boost.org] En nombre de Steven Watanabe [watanabesj(a)gmail.com] Enviado el: miércoles, 10 de septiembre de 2008 18:54 Para: boost-users(a)lists.boost.org Asunto: Re: [Boost-users] [bind] [multi-index] Create type of pointer to overloaded erase method > AMDG > > Peter Barker wrote: > > RecordsContainer::size_type (RecordsContainer::*erase)(unsigned int) = > > &RecordsContainer::erase; > > > > but it failed to compile. Visual Studio 2003 gives me the following > > error on the above line: > > <snip> > > > > The following compiles for me with msvc 9.0 > > int main() { > RecordsContainer::size_type (RecordsContainer::*erase)(unsigned int) = > &RecordsContainer::erase; > } Oddly enough, it fails with MSVC++ 8.0 (though it shouldn't since your call_traits analysis is correct): error C2440: 'initializing' : cannot convert from 'overloaded-function' to 'boost::multi_index::detail::ordered_index<KeyFromValue,Compare,SuperMeta, TagList,Category>::size_type (__thiscall boost::multi_index::multi_index_container<Value,IndexSpecifierList>::* )(unsigned int)' [...] The following is bulletproof and has the added benefit of its immunity against future slight changes in the interfaces (like, for instance, erase(key) getting templatized): template<typename Container,typename Key> typename Container::size_type container_erase(Container& c,const Key& k) { return c.erase(k); } int main() { RecordsContainer::size_type (*erase)(RecordsContainer&,const unsigned int&) = &container_erase; } Joaquín M López Muñoz Telefónica, Investigación y Desarrollo

1 0

[serialization] How to deal with bit fields?
by Sebastian Knopp 10 Sep '08

10 Sep '08

Hello, I am a new user of the serialization library and find it is a very comfortable solution. But I met a problem concerning the serialization of bit fields. For example, look at the following: struct A { int a : 1; int b : 7; }; Serialization using the following code does not work, because bitfields can not be passed by reference. template<class Archive> void serialize(Archive& ar, A& theA, const unsigned int version) { ar & make_nvp("a", theA.a); ar & make_nvp("b", theA.b); } So, I use this workaround: template<class Archive> void serialize(Archive& ar, A& theA.a, const unsigned int version) { bool myA; int myB; if (!Archive::is_loading::value) { myA = (theA.a == 1); myB = theA.b; } ar & make_nvp("a", myA); ar & make_nvp("b", myB); if (Archive::is_loading::value) { theA.a = myA ? 1 : 0; theA.b = myB; } } Do you have any suggestions how a more elegant solution could look like? Thanks and bye, Sebastian Knopp

2 1

[dataflow] Filter with multiple slots
by Manuel Jung 10 Sep '08

10 Sep '08

Hey, I want to define my own filter, which can take different signatures. It should take a Url (e.g. std::string, for me an unicode type) and use a difault timeout or it should take both. There is an tutorial for multiple slots, when using a consumer, but i need a filter and i am wondering how i should define the signature of the filter then? (overlaoding of the operator() method needs to be done of course. Kind Regards Manuel

2 1