- Boost-users - lists.boost.org

Re: [Boost-users] Serialization -- Which constructor is called?
by Eric-public 13 Jul '05

13 Jul '05

Sorry, I should have described my usage a bit better. I am de-serializing from a pointer to a class which then contains vectors of other shared_ptr's to yet more classes. None of the default constructors seem to be hitting my breakpoints -- which could be a GDB issue for all I know. I'll add the load_construct_data function to my class to see if that helps. On a side issue, I sometimes get stream_exception's thrown when de-serializing. Since the stream is still good, I'm assuming that this is caused by an un-registered class being de-serialized. Do you have any pointers as to how I can debug this? Thanks, Eric > > >Robert Ramey wrote: > >A couple of hints: > >a) constructor is called only when de-serializing a pointer. So unless >you're doing this you won't see and constructor call. > >b) When creating a new object in the course of de-serializing a pointer, the >function load_construct_data is callled. This calls an in place new for the >default constructor. By specializing this function for your particular >class, you can arrange to call a non-default constructor. > >Robert Ramey > >Eric wrote: >> In debugging a problem, I put breakpoints on the default and copy >> constructor along with the operator=. When de-serializing the class, >> none of the breakpoints were hit. >> >> I've searched through the documentation and old posts and can't seem >> to find any reference to which constructor is called from the >> serialization class. Could someone enlighten me on this? I'm sure >> it's covered somewhere, but I've somehow missed it. >> >> Thanks, >> >> Eric >> >> Boost 1.32 under Linux ________________________________________________ Message sent using UebiMiau 2.7.2

1 0

How to delete all vertices and edges in BGL
by Abhijit Deshmukh 12 Jul '05

12 Jul '05

Hi, I have a mutable property undirected graph. I want to delete all the vertices and edges in the graph. I tried using the for loop for( b = *vertices( databaseGraph ).first; b!= *vertices( databaseGraph ).second; b++ ){ clear_vertex( b, databaseGraph ); remove_vertex( b, databaseGraph ); } This works fine for some extent, but then crashes. My guess is that the loop crashes when it reaches a vertex whose edges have not been deleted. But my speculation was that clear_vertex deletes all the edges associated with a vertex and the function should not crash when clear_vertex is called before remove_vertex. Clearly, I am making a mistake, but am not able to point it out. Can you let me know what is the correct way of deleting all edges and vertices in a mutable property undirected graph. Thanks in anticipation. Regards, Abhijit

4 5

internal property map - invalid key
by David Ohlemacher 12 Jul '05

12 Jul '05

Hi all, According to the docs <http://www.boost.org/libs/property_map/property_map.html>, "Invoking a property map function on an /invalid/ key results in undefined behavior." True enough; memory is corrupted. How do I prevent this if I am not sure my key is valid. Do I have to keep a list of valid keys separately from my property map? Thanks, -d #include <iostream> #include <map> #include <string> #include <boost/property_map.hpp> template <typename AddressMap> void foo(AddressMap address) { typedef typename boost::property_traits<AddressMap>::value_type value_type; typedef typename boost::property_traits<AddressMap>::key_type key_type; value_type old_address, new_address; key_type fred = "Fred"; old_address = get(address, fred); new_address = "384 Fitzpatrick Street"; put(address, fred, new_address); key_type joe = "Joe"; value_type& joes_address = address[joe]; joes_address = "325 Cushing Avenue"; } int main() { std::map<std::string, std::string> name2address; boost::associative_property_map< std::map<std::string, std::string> > address_map(name2address); name2address.insert(make_pair(std::string("Fred"), std::string("710 West 13th Street"))); name2address.insert(make_pair(std::string("Joe"), std::string("710 West 13th Street"))); foo(address_map); for (std::map<std::string, std::string>::iterator i = name2address.begin(); i != name2address.end(); ++i) std::cout << i->first << ": " << i->second << "\n"; return EXIT_SUCCESS; }

1 0

Re: [Boost-users] Problem with boost::filesystem::copy_file
by alice.johnson＠sophos.com 12 Jul '05

12 Jul '05

On 02/07/2005 02:58:17 <bdawes(a)acm.org> wrote: ><alice.johnson(a)sophos.com> wrote in message >news:OF8E666308.D3068D40-ON80257031.0048D4C9-80257031.00495 D06(a)sophos.com... >> Hi, >> >> I think I've found a problem with this function. On Windows 98/Me >> computers, it can fail with path not found error (3) if the from path is >> on >> a network share. > >*&%#! and several other words not fit for a public mailing list. If you ever >hear I've gone crazy, blame differing Windows versions. Although I'm reading >a book that tries to explain the differences between Unix flavors, and that >is a mess too. > >> I would suggest the following change to >> src\filesystem\operations_posix_windows.cpp: >> >> Currently: >> >> if ( !::CopyFileA( from_file_ph.string().c_str(), >> to_file_ph.string().c_str(), /*fail_if_exists=*/true >> ) ) >> Suggested change: >> >> if ( !::CopyFileA( from_file_ph.native_file_string().c_str(), >> to_file_ph.native_file_string().c_str(), >> /*fail_if_exists=*/true ) ) >> >> There may be other areas in the filesystem code where it would be better >> to >> use native_file_string(). > >It may be that all uses should be changed. I'm a bit worried about doing >that so close to the 1.33 release. Not to mention that that code is about to >be retired anyhow. > >Let me sleep on it. I've also got to find a machine to test on; I never >owned a copy of Win 98/me. So why is this code to be retired? Will there be an alternative? <janpstetka(a)supanet.com> wrote: >Is not trying to support 98/ME an option? No. But I have my own work around; I don't need a fix. Just thought I'd let you know about the problem. Alice

1 0

String Tokenizer question
by J Liu 12 Jul '05

12 Jul '05

I am using boost 1.31 Tokenizer. The question I have is Can boost tokenizer change the predicates on the fly? For example: string s = "my birthday is 1976, 15/17"; boost::tokenizer<> tok(s); boost::tokenizer<>::iterator beg=tok.begin(); // here *beg will give us "my" typedef boost::tokenizer<boost::char_separator<char> > Boost_char_tokenizer; boost::char_separator<char> sep(",/"); Boost_char_tokenizer next(s, sep); Boost_char_tokenizer::iterator tok_iter = next.begin(); for(; tok_iter!=next.end(); ++tok_iter){ std::cout << *tok_iter << std::endl; } // this will give us "my birthday is 1976", "15" and "17" But what I want to get is "my", "birthday is 1976", "15" and "17". In other words, I want the tokenizer either can give us the remaining of the string or can change the delimiters dynamically. Is there a way to do this? Thanks for any help.

2 2

Re: [Boost-users] String Tokenizer question
by Bjorn.Karlsson＠readsoft.com 12 Jul '05

12 Jul '05

> But what I want to get is "my", "birthday is 1976", "15" and "17". > In other words, I want the tokenizer either can give us the remaining of the > string or can change the delimiters dynamically. > Is there a way to do this? You need to use two tokenizers, just like in the example you provided. The only thing you need to do differently is retrieve an iterator to the current position in the input string after the first tokenizer iterator (beg) has been dereferenced, and pass the iterator range [current,end) to the constructor of the tokenizer next rather than the whole string (s). You can retrieve the current iterator from a token_iterator by calling the iterator's member function base(). Here's your example again, modified to do just that: std::string s = "my birthday is 1976, 15/17"; boost::tokenizer<> tok(s); boost::tokenizer<>::iterator beg=tok.begin(); // here *beg will give us "my" typedef boost::tokenizer<boost::char_separator<char> > Boost_char_tokenizer; boost::char_separator<char> sep(",/"); Boost_char_tokenizer next(beg.base(),s.end(), sep); Boost_char_tokenizer::iterator tok_iter = next.begin(); for(; tok_iter!=next.end(); ++tok_iter){ std::cout << *tok_iter << std::endl; } // this will give us " birthday is 1976", " 15" and "17" Note that the only difference in this version is the construction of next: Boost_char_tokenizer next(beg.base(),s.end(), sep); Bjorn

2 1

How to serialize a derived class object with base class pointer?
by Yan Zhang 11 Jul '05

11 Jul '05

Hi, I'm trying to serialize an object of derived class with a base class pointer, somehow only base class's serialization member function got called. Here is the code (simplified): class CMyCls1 { public: CMyCls1() : m(10) {} ~CMyCls1() {} private: friend class boost::serialization::access; template<class Archive> void serialize(Archive &ar, const unsigned int version) { ar & BOOST_SERIALIZATION_NVP(m); } int m; }; class CMyCls2 : public CMyCls1 { public: CMyCls2() : x(1.5) {} ~CMyCls2() {} private: friend class boost::serialization::access; template<class Archive> void serialize(Archive &ar, const unsigned int version) { ar & BOOST_SERIALIZATION_BASE_OBJECT_NVP(CMyCls1); ar & BOOST_SERIALIZATION_NVP(x); } double x; }; void main () { CMyCls1 *pmycls = new CMyCls2; ... oa.register_type(static_cast<CMyCls1 *>(NULL)); oa.register_type(static_cast<CMyCls2 *>(NULL)); oa << BOOST_SERIALIZATION_NVP(pmycls); ... } Look at the archive file, only base class data is serialized. Actuall, CMyCls2::serialization() is not called at all. How do you serialize a derived class object with a pointer of base class? Thanks, Yan

2 1

Problem with Graphviz (read) and subgraphs
by odisseo＠despammed.com 11 Jul '05

11 Jul '05

Can you help me writing a simple function that iterates on all subgraphs in the .dot file and print out the 1. subgraph name 2. for each edge source label / edge label / target label ? I've tryed with this but didn't work: [...] boost::GraphvizDigraph g; boost::read_graphviz(efsmfile, g); [...] boost::GraphvizDigraph::children_iterator subgraph_index, subgraph_end; for (boost::tie(subgraph_index, subgraph_end) = g.children(); subgraph_index != subgraph_end; subgraph_index++) { cout << get_graph_name(*subgraph_index) << endl; } const string& EFSM::get_graph_name(const Graph& g) { const typename boost::graph_property<Graph, boost::graph_name_t>::type& name = boost::get_property(g, boost::graph_name); return name; }

1 0

[variant] not copy-constructable
by Felipe Magno de Almeida 11 Jul '05

11 Jul '05

How can I work this out? I've tried with auto_ptr<object> but its copy-constructor is explicit and didnt worked anyway. (MSVC71). What I wanted is a vector of variant<object1, object2> and I dont want object1 and object2 to be copied because it is expensive. But Im not figuring out how to do this... tried boost::ref either with no success... :/ Any information is appreciated, thanks in advance, Felipe. -- Felipe Magno de Almeida Developer from synergy and Computer Science student from State University of Campinas(UNICAMP). Unicamp: http://www.ic.unicamp.br Synergy: http://www.synergy.com.br "There is no dark side of the moon really. Matter of fact it's all dark."

1 0

[spirit] help with phoenix.
by Zeljko Vrba 11 Jul '05

11 Jul '05

Hi! I'm trying to create a lisp-like 'map' functor. however, gcc complains with (..follows). It seems that somehow it has a wrong idea of the return value of the functor return value. The complete code (the header and the test program) are in the attachment. stage.h: In member function `typename phoenix::actor_result<map_c<ProviderT, OperationT>, TupleT>::type map_c<ProviderT, OperationT>::eval(const TupleT&) const [with TupleT = phoenix::tuple<phoenix::nil_t, phoenix::nil_t, phoenix::nil_t, phoenix::nil_t>, ProviderT = phoenix::actor<phoenix::composite<typed_binary_istream_f<unsigned int>, phoenix::nil_t, phoenix::nil_t, phoenix::nil_t, phoenix::nil_t> >, OperationT = phoenix::actor<phoenix::composite<phoenix::times_op, phoenix::actor<phoenix::composite<phoenix::dereference_op, phoenix::actor<phoenix::argument<0> >, phoenix::nil_t, phoenix::nil_t, phoenix::nil_t> >, phoenix::actor<phoenix::value<int> >, phoenix::nil_t, phoenix::nil_t> >]': /usr/pkg/include/boost/spirit/phoenix/actor.hpp:354: instantiated from `typename phoenix::actor_result<BaseT, phoenix::tuple<phoenix::nil_t, phoenix::nil_t, phoenix::nil_t, phoenix::nil_t> >::type phoenix::actor<BaseT>::operator()() const [with BaseT = map_c<phoenix::actor<phoenix::composite<typed_binary_istream_f<unsigned int>, phoenix::nil_t, phoenix::nil_t, phoenix::nil_t, phoenix::nil_t> >, phoenix::actor<phoenix::composite<phoenix::times_op, phoenix::actor<phoenix::composite<phoenix::dereference_op, phoenix::actor<phoenix::argument<0> >, phoenix::nil_t, pho enix::nil_t, phoenix::nil_t> >, phoenix::actor<phoenix::value<int> >, phoenix::nil_t, phoenix::nil_t> > >]' stage_test04.cc:34: instantiated from here stage.h:172: error: conversion from `const int' to non-scalar type `const phoenix::nil_t' requested #ifndef STAGE_H__ #define STAGE_H__ #include <utility> #include <istream> #include <boost/spirit/phoenix/functions.hpp> #include <boost/spirit/phoenix/composite.hpp> /** * @file * Stage fundamentals. Each stage returns a POINTER to value. 0 means that * the stage is exhausted, and 0 will be returned each time afterwards. * * Some conventions: _p are primitives, _c are composites, _f functions. */ /***************************************************************************** * Typed binary stream reads items from an istream and returns a pointer to * the newly read item. When there are no more items to read, returns 0. ****************************************************************************/ template<class ItemT> struct typed_binary_istream_f { typedef ItemT *result_type; ItemT *operator()() const { return is.read((char*)&storage, sizeof(storage)) ? const_cast<ItemT*>(&storage) : 0; } typed_binary_istream_f(std::istream &is_) : is(is_) { } std::istream &is; ItemT storage; }; template<class ItemT> phoenix::function<typed_binary_istream_f<ItemT> > typed_binary_istream(std::istream &is) { return typed_binary_istream_f<ItemT>(is); } /***************************************************************************** * Calls the provider repeatedly until it returns 0. Operation is executed on * all elements returned by the provider. ****************************************************************************/ template<typename ProviderT, typename OperationT> struct until_eof_c { typedef until_eof_c<ProviderT, OperationT> self_t; template<typename TupleT> struct result { typedef void type; }; until_eof_c(ProviderT const &provider_, OperationT const &op_) : provider(provider_), operation(op_) { } template<typename TupleT> void eval(TupleT const &args) const; ProviderT provider; OperationT operation; }; template<typename ProviderT, typename OperationT> template<typename TupleT> void until_eof_c<ProviderT, OperationT>::eval(TupleT const &args) const { typename phoenix::actor_result<ProviderT, TupleT>::type p; while((p = provider())) operation(p); } template<typename ProviderT, typename OperationT> phoenix::actor<until_eof_c< typename phoenix::as_actor<ProviderT>::type, typename phoenix::as_actor<OperationT>::type> > until_eof(ProviderT const &provider, OperationT const &operation) { typedef until_eof_c< typename phoenix::as_actor<ProviderT>::type, typename phoenix::as_actor<OperationT>::type> result; return result( phoenix::as_actor<ProviderT>::convert(provider), phoenix::as_actor<OperationT>::convert(operation)); } /***************************************************************************** * Returns only elements that match some predicate. ****************************************************************************/ template<typename ProviderT, typename PredicateT> struct filter_c { typedef filter_c<ProviderT, PredicateT> self_t; template<typename TupleT> struct result { typedef typename phoenix::actor_result<ProviderT, TupleT>::type type; }; filter_c(ProviderT const &provider_, PredicateT const &p_) : provider(provider_), predicate(p_) { } template<typename TupleT> typename phoenix::actor_result<self_t, TupleT>::type eval(TupleT const &args) const; ProviderT provider; PredicateT predicate; }; template<typename ProviderT, typename PredicateT> template<typename TupleT> typename phoenix::actor_result< typename filter_c<ProviderT, PredicateT>::self_t, TupleT>::type filter_c<ProviderT, PredicateT>::eval(TupleT const &args) const { typename phoenix::actor_result<ProviderT, TupleT>::type p; while((p = provider())) if(predicate(p)) return p; return 0; } template<typename ProviderT, typename PredicateT> phoenix::actor<filter_c< typename phoenix::as_actor<ProviderT>::type, typename phoenix::as_actor<PredicateT>::type> > filter(ProviderT const &provider, PredicateT const &operation) { typedef filter_c< typename phoenix::as_actor<ProviderT>::type, typename phoenix::as_actor<PredicateT>::type> result; return result( phoenix::as_actor<ProviderT>::convert(provider), phoenix::as_actor<PredicateT>::convert(operation)); } /***************************************************************************** * Apply operation to each element of the provider and return the transformed * sequence. This is called 'map' in LISPs. ****************************************************************************/ template<typename ProviderT, typename OperationT> struct map_c { typedef map_c<ProviderT, OperationT> self_t; template<typename TupleT> struct result { typedef typename phoenix::actor_result<OperationT, TupleT>::type type; }; map_c(ProviderT const &provider_, OperationT const &op_) : provider(provider_), operation(op_) { } template<typename TupleT> typename phoenix::actor_result<self_t, TupleT>::type eval(TupleT const &args) const; ProviderT provider; OperationT operation; }; template<typename ProviderT, typename OperationT> template<typename TupleT> typename phoenix::actor_result< typename map_c<ProviderT, OperationT>::self_t, TupleT>::type map_c<ProviderT, OperationT>::eval(TupleT const &args) const { typename phoenix::actor_result<ProviderT, TupleT>::type p; p = provider(); // while((p = provider())) // if(operation(p)) return p; return operation(p); } template<typename ProviderT, typename OperationT> phoenix::actor<map_c< typename phoenix::as_actor<ProviderT>::type, typename phoenix::as_actor<OperationT>::type> > map(ProviderT const &provider, OperationT const &operation) { typedef map_c< typename phoenix::as_actor<ProviderT>::type, typename phoenix::as_actor<OperationT>::type> result; return result( phoenix::as_actor<ProviderT>::convert(provider), phoenix::as_actor<OperationT>::convert(operation)); } #ifdef STAGE_INSTANTIATE__ #endif #endif // STAGE_H__ #include <iostream> #include <fstream> #include <boost/spirit/phoenix/primitives.hpp> #include <boost/spirit/phoenix/operators.hpp> #define STAGE_INSTANTIATE__ #include "stage.h" using namespace std; using namespace phoenix; struct print_f { template<typename ItemT> struct result { typedef bool type; }; template<typename ItemT> bool operator()(ItemT const &item) const { cout << item << endl; return item; } }; static function<print_f> print; int main(int argc, char **argv) { unsigned int *item; fstream f("test1.bin"); if(!f) { cerr << "can't open test1.bin\n"; return 1; } map(typed_binary_istream<unsigned int>(f)(), *arg1 * 2) (); return 0; }

1 0