- Boost-users - lists.boost.org

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

Re: [Boost-users] [general] MSVC71 Warnings Off?
by Wu Yinghui, Freddie 16 Aug '06

16 Aug '06

> Hi Freddie. > >> Thanks for your quick reply. I knew this trick. But what I want is >> probably some more elegant ways of doing this. (It seems quite a bit of >> a trouble since those pragmas will generate lots of warnings when I >> build using gcc. And if I have to surround each of those pragma with >> platform detection macros, it will end up quite a few lines for each of >> my hundred header/source files.) >> >> Any easier-to-adopt ideas? :) > > Hmmm... I do not think I explained myself clearly enough the last > time. :-) > > If you look at the header files I sent you they already have platform > detection guards in them. And the whole point of those headers is to > have them and only them contain the platform detection and the > necessary pragmas. After that you just use those BoostWrappers > headers instead of the original ones from boost. That means that you > add NO EXTRA LINES to any of your own hundred header/source > files. :-) > > I also told you that we at first had the same fears as you, but as it > turns > out actually very few boost libraries cause warnings and so there are > actually very few of those BoostWrappers headers. Look the archive > I attached in the last message - there are less maybe 10-20 files in there > and their contents is very stable. Adding new ones is rare and when it > does happen it happens natually - compiling your other code tells you > directly which warnings to disable and in which header files. > > Hope this clears things up... :-) > > Best regards, > Jurko Gospodnetic Sure it does. I guess this is one of the best approach I can see for now. But I had problem getting the attachment you mentioned from your previous post. (I didn't see any attachment.) Is it possible that you send the header files again? Sorry for the trouble. Cheers, Freddie

1 0

Re: [Boost-users] Problem building for VC-8
by Bob Cowdery 16 Aug '06

16 Aug '06

Johan I did get a build by copying the msvc vc8 extension text into a file vc8-tools.jam and then using -sTOOLS=vc8 on the command line. I don't know if this is the correct way or not but it seemed to work. Bob -----Original Message----- From: boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org]On Behalf Of Johan Nilsson Sent: 16 August 2006 07:36 To: boost-users(a)lists.boost.org Subject: Re: [Boost-users] Problem building for VC-8 Bob Cowdery wrote: > Hi > > I am trying to build the boost libraries on XP with the VC-8 compiler > from the Express edition. I have downloaded boost-1-33-1 and bjam > boost-jam-3.1.13-1-ntx86. I tried to follow the instructions at > http://www.boost.org/tools/build/v1/vc-8_0-tools.html. It's not clear > to me if the msvc tools and the vc8 extension have to be installed or > are part of bjam. I can't find a download for the tools though. > Ignoring that, I simply ran up the hopefully correct environment by > running "C:\Program Files\Microsoft Platform SDK for Windows Server > 2003 R2\Setup\SetEnvLaunchWinXP32Retail.Cmd" /XP32 /RETAIL and then > vcvars32.bat. I also set VC80_ROOT to my VC8 directory. > > Whatever I try I get the same result as below. It's trying to invoke > VC7 which I don't have. > > Any help appreciated. This isn't really a reply to your question, but did you try the installer at: http://boost-consulting.com/download.html // Johan _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

2 1

Re: [Boost-users] [tuple] generating tuples from type lists
by imre＠u-tx.com 16 Aug '06

16 Aug '06

From: Joel de Guzman <joel(a)boost-consulting.com> > > Oh. You are looking at Fusion1. That was the pre-review version. > We now have Fusion2 (which I will put in the CVS soon) which is > the review version. Sorry for the confusion. I played a bit with this pre review wersion (which is accidentaly in the released boost version 1.33.1), and I sort of getting stuck when I am trying to put arrays into the tuple with make_tuple(). Is there a way to do it? Thx ImRe

2 1

Re: [Boost-users] [tuple] generating tuples from type lists
by imre＠u-tx.com 16 Aug '06

16 Aug '06

From: David Abrahams <dave(a)boost-consulting.com> > Is it possible to do this trick during compilation time? > I was able to go from type lists to boost::cons but I can't find a way > to make that last step. Not sure what you mean by "that last step..." namespace mpl = boost::mpl; using namespace mpl::placeholders; typedef mpl::vector<int, long, char const*> types; typedef mpl::fold<types, boost::cons<_2,_1> >::type tuple_type; tuple_type some_tuple; When I try to compile this, I receive the following error: boost/mpl/aux_/preprocessed/plain/apply_wrap.hpp(48) : error C2039: 'apply' : is not a member of 'boost::mpl::na' What does this means? Thx ImRe

1 0