August 2006 - Boost-users - lists.boost.org

[statechart] Getting the current state from an async machine
by Johan Nilsson 17 Aug '06

17 Aug '06

Hi, I need a way to get the current state from an asynchronous state machine for testing purposes - something analogous to state_cast. I'll only be using this in my unit tests, so I don't mind if it's a bit cumbersome to use. Thread-safety also isn't an issue. Is there any way of getting a reference to the actual Processor instance? Thanks / Johan

2 9

[serialization] -- compiling but not linking...undefined references
by rkemp 17 Aug '06

17 Aug '06

I successfully built the entire boost library using the bjam utility. Now I'm trying to compile and link demo.cpp but it is failing to resolve about a dozen symbols. demo.cpp is the example code for the serialization library. A typical error reads: undefined reference to 'boost::archive::ARCHIVE_SIGNUTURE' I am linking to the libboost_serialization-gcc.so library. Is there another? Can anyone steer me in the right direction? Thx.

2 1

[program_options] How to allow multiple options?
by Ian McCulloch 17 Aug '06

17 Aug '06

Hi, I have spent several hours looking at the docs and sources, but I cannot figure out how to allow multiple occurances of an option. Many unix tools allow you to specify the --verbose option multiple times, to mean increasing verbosity of output, and I want to replicate this behaviour. Does this require writing a new class derived from value_semantic, that combines the behaviour of bool_switch() and typed_value<std::vector<bool>> ? I'm not sure how to go about doing that. It is a very useful lib, but I do wish (often!) there was some real reference docs (aside from the doxygen stuff and comments in the sources themselves, which isn't enough :-( Unrelated, but in trying to figure this out from the docs, I noticed in the overview.html, in the Syntactic Information there is a call to a function value<string>()->implicit(). However typed_value<T> has no such member function. Maybe it is left over from a previous design iteration or something, as there is a member variable m_implicit but it is never initialized or accessed. Cheers, Ian McCulloch

2 2

[serialization] Problem serializing a std::map
by David Guardado 16 Aug '06

16 Aug '06

Hello, When trying to serialize a STL map, doing the following (in Visual C++ 7.1): class Being { private: std::wstring Name, Description; friend class boost::serialization::access; public: template<class Archive> void serialize(Archive & ar, const unsigned int version) { ar & this->Name; ar & this->Description; } }; class Ecosystem { private: friend class boost::serialization::access; std::map<std::wstring,Being> Beings; public: template<class Archive> void serialize(Archive & ar, const unsigned int version) { ar & this->Beings; } }; ... ar & ecosystem; // Making the serialization within some function (storing) ... I get this compile-time error: "c:\Boost\include\boost-1_33_1\boost\serialization\access.hpp(109): error C2039: 'serialize' : is not a member of 'std::map<_Kty,_Ty>' with [_Kty=std::wstring,_Ty=Models::Beings]" Has anyone any ideas about the reason for this error? Thanking you in advance, David

2 2

Re: [Boost-users] random number generator 160 bit.
by james.jones＠firstinvestors.com 16 Aug '06

16 Aug '06

From: "chun ping wang" <cablepuff(a)gmail.com> > hmm whats the best [arbitrary-length integer class]? I really don't know which one is best. I recommend looking at several, reading the docs, and trying out the ones that look promising. - James Jones Administrative Data Mgmt. Webmaster 375 Raritan Center Pkwy, Suite A Data Architect Edison, NJ 08837

1 0

[Graph] kamada_kawai_spring_layout runs forever
by Sebi 16 Aug '06

16 Aug '06

hello, i'm trying to use the boost graph libarary to layout a small (< 10 nodes), fully connected, undirected graph. the problem is, that the kamada_kawai_spring_layout() function never returns. when i comment out the code to set the weights for my graph it works, but then it is of course not of much use for me. below is the sourcecode i came up with. as i said, when i comment out the line put(weightMap, e, 1.0); it works (i.e. the program finishes after a short time) and all nodes are at the same location i think the problem could be the way i connect the nodes, which is quite clumsy here. maybe someone has a better idea how to accomplish this. regards and thanks in advance, sebastian mecklenburg #include "boost/graph/adjacency_list.hpp" #include "boost/graph/graph_traits.hpp" #include "boost/graph/kamada_kawai_spring_layout.hpp" #include <iostream> using namespace boost; enum vertex_position_t { vertex_position }; namespace boost { BOOST_INSTALL_PROPERTY(vertex, position); } struct pos { double x; double y; }; typedef adjacency_list<listS, listS, undirectedS, // Vertex properties boost::property<boost::vertex_index_t, int, boost::property<vertex_position_t, pos> >, // Edge properties property<edge_weight_t, double> > WUGraph; typedef graph_traits<WUGraph>::vertex_descriptor vertex_descriptor; typedef graph_traits<WUGraph>::edge_descriptor edge_descriptor; int main(int argc, char *argv[]) { int numNodes = 5; WUGraph ug; for(int i = 0; i < numNodes; ++i) { vertex_descriptor u = boost::add_vertex(ug); boost::put(vertex_index, ug, u, i); } graph_traits<WUGraph>::vertex_iterator vi, vi_end; graph_traits<WUGraph>::vertex_iterator vj, vj_end; boost::property_map<WUGraph, vertex_index_t>::type indexMap = get (vertex_index, ug); boost::property_map<WUGraph, edge_weight_t>::type weightMap = get (edge_weight, ug); for (tie(vi, vi_end) = vertices(ug); vi != vi_end; ++vi) { for (tie(vj, vj_end) = vertices(ug); vj != vj_end; ++vj) { int i1 = get(indexMap, *vi); int i2 = get(indexMap, *vj); if(i1 < i2) { edge_descriptor e = add_edge(*vi, *vj, ug).first; put(weightMap, e, 1.0); } } } boost::kamada_kawai_spring_layout(ug, get(vertex_position, ug), get(edge_weight, ug), side_length(50.0)); boost::property_map<WUGraph, vertex_position_t>::type positionMap = get(vertex_position, ug); for (tie(vi, vi_end) = vertices(ug); vi != vi_end; ++vi) { std::cout << " n" << get(vertex_index, ug, *vi) << "[ pos=\"" << (int)positionMap[*vi].x + 25 << ", " << (int)positionMap [*vi].y + 25 << "\" ];\n"; } }

2 1

[serialization] best practice for boost::serialization + BerkeleyDB?
by Andy Schlaikjer 16 Aug '06

16 Aug '06

I'm interested in using boost::serialization for some small classes which need to be backed by a database. I thought I'd try and take advantage of boost::serialization to manage the basic marshaling of objects into/out of a binary format, and use BerkeleyDB for database support. Unfortunately, I looked around for some prior art, but couldn't find any, so I'm posting here a small example of my own with the hope that others may lend feedback on this basic approach, or else find use for the ideas here in their own work. The following code was built and (minimally) tested with gcc-3.4.6, stlport-5.0.2, berkeleydb-4.4.20 (with stlport), boost-1.33.1 (with gcc/stlport). Thanks! Andy <code> #include <string> #include <iostream> #include <sstream> #include <boost/archive/binary_oarchive.hpp> #include <boost/archive/binary_iarchive.hpp> #include <db_cxx.h> using namespace std; /** * An overly simplistic representation of an every-day schmo. */ class Person { private: unsigned int id_; // a horrible reduction /** * Inserts a textual representation of a Person into an output stream (not * intended for serialization). * @param os an output stream * @param p a Person to insert into the output stream * @return the output stream */ friend ostream & operator<<(ostream & os, Person const & p) { return os << p.id_; } friend class boost::serialization::access; friend class Database; /** * Serializes this Person into/out-of the given Archive, depending on the type * of the Archive. * @param ar an input or output Archive * @param version a class number */ template<typename Archive> void serialize(Archive & ar, unsigned int const version) { ar & id_; } /** * Serializes this Person into the given Dbt. * @param[out] data the Dbt into which to serialize this Person */ void serialize(Dbt * data) const { // serialize this Person into binary archive ostringstream oss(ios::out | ios::binary); boost::archive::binary_oarchive oa(oss, boost::archive::no_header); oa << (*this); // copy serialized data (is there any way to avoid this?) void * buf = malloc(sizeof(char) * oss.str().size()); memcpy(buf, oss.str().data(), oss.str().size()); // initialize Dbt* memset(data, 0, sizeof(Dbt)); data->set_data(buf); data->set_size(oss.str().size()); data->set_ulen(oss.str().size()); data->set_flags(DB_DBT_USERMEM); } /** * Constructs a Person from the given Dbt key-data pair. * @param[in] key a primary key containing a name (ignored for now) * @param[in] data an element which references a serialized Person */ Person(Dbt const * key, Dbt const * data) { // construct input stream from data's void * istringstream iss (string(reinterpret_cast<char const *>(data->get_data()), data->get_size()), ios::in | ios::binary); boost::archive::binary_iarchive ia(iss, boost::archive::no_header); ia >> (*this); } /** * Extracts a secondary key (a Person's Id) from a given key-data * (name-Person) pair. * @param[in] secondary the secondary Db * @param[in] key the primary key, containing a Person name * @param[in] data the Person associated with the primary key * @param[out] skey the secondary key to be generated * @return zero on successful generation of secondary key, non-zero on any * error so BDB will ignore the possibly bogus secondary key. */ static int getSecondaryKey(Db * secondary, Dbt const * key, Dbt const * data, Dbt * skey) { unsigned int * id = reinterpret_cast<unsigned int *>(malloc(sizeof(unsigned int))); istringstream iss (string(reinterpret_cast<char const *>(data->get_data()), data->get_size()), ios::in | ios::binary); boost::archive::binary_iarchive ia(iss, boost::archive::no_header); ia >> (*id); memset(skey, 0, sizeof(Dbt)); skey->set_data(id); skey->set_size(sizeof(unsigned int)); skey->set_ulen(sizeof(unsigned int)); skey->set_flags(DB_DBT_APPMALLOC); return 0; } public: /** * Constructs a Person with the specified Id. */ Person(unsigned int id = 0) : id_(id) {} /** * Destroys this Person */ ~Person() {} /** * @return the Id of this Person */ unsigned int getId() { return id_; } }; BOOST_CLASS_IMPLEMENTATION(Person, boost::serialization::object_serializable) BOOST_CLASS_TRACKING(Person, boost::serialization::track_never) /** * A helper class to manage a pair of BerkeleyDB instances (a primary and an * associated secondary database) used for storing Persons. The primary db is * keyed on a string name, while the secondary index is keyed on Person Id. */ class Database { private: Db db_primary_; Db db_secondary_; /** * Default ctor. we hide this. */ Database() : db_primary_(NULL, 0), db_secondary_(NULL, 0) {} /** * Closes Db instances in proper order. Called by dtor. */ void close() { db_secondary_.close(0); db_primary_.close(0); } public: /** * Constructs a Database using the given file to store both primary and * secondary databases. * @param[in] filename a pathname to a file in which to store data */ Database(string const & filename) : db_primary_(NULL, 0), db_secondary_(NULL, 0) { db_primary_.open (0, // transaction id filename.c_str(), // file name "primary", // database name DB_BTREE, // database type DB_CREATE, // flags 0); // mode (see chmod(2)) db_secondary_.open (0, filename.c_str(), "secondary", DB_BTREE, DB_CREATE, 0); // associate the secondary with the primary db_primary_.associate (0, // transaction id &db_secondary_, // secondary db &Person::getSecondaryKey, // callback DB_IMMUTABLE_KEY); // flags } /** * Destroys this Database */ ~Database() { close(); } /** * Stores a Person in this Database using the specified name as a primary key. * @param[in] name a string name to use as primary key for the given Person * @param[in] p the Person to insert into this Database */ void store(string const & name, Person const & p) { Dbt key, data; key.set_data (reinterpret_cast<void *> (const_cast<char *> (name.data()))); key.set_size(sizeof(char) * name.size()); p.serialize(&data); db_primary_.put(0, &key, &data, 0); } /** * Loads a Person from this Database using the specified name as a primary key. * @param[in] name the string name to use as primary key * @return a newly constructed Person (caller is responsible for deleting this * Person) */ Person * load(string const & name) { Dbt key, data; key.set_data (reinterpret_cast<void *> (const_cast<char *> (name.data()))); key.set_size(sizeof(char) * name.size()); db_primary_.get(0, &key, &data, 0); return new Person(&key, &data); } /** * Loads a Person from this Database using the specified Id as a secondary key. * @param[in] id the Id to use as secondary key * @return a newly constructed Person (caller is responsible for deleting this * Person) */ Person * load(unsigned int id) { Dbt key, pkey, data; key.set_data(reinterpret_cast<void *>(&id)); key.set_size(sizeof(unsigned int)); db_secondary_.pget(0, &key, &pkey, &data, 0); return new Person(&pkey, &data); } }; /** * Performs a minimal test of the above Person database and serialization * facilities. */ int main(int argc, char * args[]) { cout << "Opening database" << endl; Database db("person.db"); cout << "Creating person" << endl; string name("Andy"); unsigned int id = 123456789; Person p(id); cout << "Storing person " << p << " using primary key \"" << name << "\"" << endl; db.store(name, p); cout << "Loading person " << p << " using secondary key \"" << id << "\"" << endl; Person * p2 = db.load(id); cout << "Loaded person " << *p2 << endl; delete p2; } </code>

3 3

[mpl] include order matters for set and map
by Yuval Ronen 16 Aug '06

16 Aug '06

The following simple code: #include <boost/mpl/map.hpp> #include <boost/mpl/set.hpp> #include <boost/mpl/pair.hpp> using namespace boost; int main(int argc, char* argv[]) { typedef mpl::map< mpl::pair<int, double> > my_map; return 0; } compiles fine. But if swap the order of #include <boost/mpl/map.hpp> and #include <boost/mpl/set.hpp> then it doesn't compile any more (MSVC 7.1). Yuval

2 2

1.34.0 release date?
by james.jones＠firstinvestors.com 16 Aug '06

16 Aug '06

Does anyone know when 1.34.0 is going to be released? Even approximately? - James Jones Administrative Data Mgmt. Webmaster 375 Raritan Center Pkwy, Suite A Data Architect Edison, NJ 08837

7 10

Re: [Boost-users] random number generator 160 bit.
by james.jones＠firstinvestors.com 16 Aug '06

16 Aug '06

From: "chun ping wang" <cablepuff(a)gmail.com> > yeah i get this error. bad numeric conversion: positive overflow Still i > need some way to generate a 160 bit integer somehow. ... :( You can write your own integer class or search Google or Sourceforge for arbitrary-sized integer implementations. Luckily, you're working in a language with templates, so once you find an implementation, just use it for your template parameter and you should be good to go. - James Jones Administrative Data Mgmt. Webmaster 375 Raritan Center Pkwy, Suite A Data Architect Edison, NJ 08837

2 1