August 2009 - Boost-users - lists.preview.boost.org

[boost-users]Shared_ptr and std::find
by Ryan McConnehey 07 Aug '09

07 Aug '09

Is it possible to have std::find compare the class that the shared_ptr is pointing too? I have a list of shared_ptr of a class called FundamentalTypes. The class has it's own operator==. I'd like to search the list with std::find using the class's operator==. I've included the class with the key elements I'm referring too. Ryan class FundamentalTypes { public: FundamentalTypes(std::string const& name) : m_name(name) bool operator==(FundamentalTypes const& rhs) { return (rhs.m_name == m_name) ? true : false; } std::string m_name; }; template<typename T> class Type : public FundamentalTypes { Type(std::string const& name) : FundamentalTypes(name) }; //main typedef shared_ptr<FundamentalTypes> sharedTypes typedef std::list<sharedTypes> TypesList sharedTypes value = sharedTypes(new Type<short>("bob")); TypesList m_list; TypesList::iterator itor = std::find(m_list.begin(), m_list.end(), value);

3 3

Re: [Boost-users] [interprocess]files locations for mutex and named shm
by Mike Spertus 07 Aug '09

07 Aug '09

I'm being bitten by the surprise discovery that different Windows users can't(?) communicate via named shared memory using Boost Interprocess as in the post below from 2008-09-18 (Changing all processes to use the same TMP directory is not an option for us due to interoperability requirements). Is there a workaround? Was anything ever done about the proposed solution below? Another alternative would be to make it possible to specify the "Boost Interprocess temp directory". Mike Eugen Stoianovici wrote: > > Ion Gaztañaga wrote: > > Eugen Stoianovici wrote: > >> Ion Gaztañaga wrote: > >>> You can't, unless you modify the code in: > >>> > >>> boost/interprocess/detail/temp_dir_helper.hpp > >>> > >>> Why do you want to change the location of those files? > >>> > >>> Ion > >> This is only a problem if you run processes with different users. Ex: > >> if you're using shared memory between an user app and a service, the > >> service will look for the segment in C:\Windows\Temp\ and the user > >> app will look for it in C:\Documents and settings\user\local > >> settings\etc. > >> My changes should go in the temp_dir_helpers.hpp file, right? > > > > Good point. I'll try to see if I can unify this when I have a bit of > > time. Maybe we can find a common directory for all processes. Any hint > > or suggestion? > > > On XP C:\Windows\Temp would be a good idea but i don't know if it is > writable by all users. I'm going to look into it when i get the chance, > but for now, changing the "TMP" env to something else is enough for me. > > Thanks again, > Eugen Stoianovici

2 1

[Serialization] Zero-sized arrays
by Roland Kaminski 07 Aug '09

07 Aug '09

Hi, i am planning to use the boost serialization library to implement checkpointing for an ASP/Sat solver. Unfortunately, the solver implements one (very special) optimization that makes use of a zero-sized array at the end of a class. I have found no way to store something like this using the library. There is a small program in the attachment that contains data structures similar to the ones used in the solver code. This would need a mechanism to control the allocation of memory during deserialization. Do you see any possibility do serialize something like this? Regards and thanks, Roland

3 11

Parsing Sparse Matrix to Get JA, IA, A Vector
by Peverall Dubois 07 Aug '09

07 Aug '09

Does boost has a library to parse a sparse matrix to get the index the non-zero element value? E.g. given this matrix: 1 2 0 0 0 3 9 0 0 1 4 0 We hope it to return A = [ 1 2 3 9 1 4 ] IA = [ 1 3 5 7 ] JA = [ 1 2 2 3 2 3 ] Let NNZ denote the number of nonzero entries of Matrix. The first array is A, which is of length NNZ, and holds all nonzero entries of M in left-to-right top-to-bottom order. The second array is IA, which is of length m + 1 (i.e., one entry per row, plus one). IA(i) contains the index in A of the first nonzero element of row i. Row i of the original matrix extends from A(IA(i)) to A(IA(i+1)-1). The third array, JA, contains the column index of each element of A, so it also is of length NNZ. Is there any implementation for handling such a large matrix to give us A, IA and JA, without having have to slurp all the files into RAM? Because the actual dimension is very large 10^6 to 10^7. Regards, G.V.

2 1

Inconsistent behavior of fusion::find with vector and adapted struct.
by Mike Tegtmeyer 06 Aug '09

06 Aug '09

I seem to have found an inconsistency in how fusion::find works with fusion::vector and an adapted struct. Basically, attempting to find a non-const type on a const vector with a non-const member works but requires a const-member find for an adapted struct. (See code below). Is this something that was just overlooked or there a technical reason for this behavior. In any case is there some way to identify a adapted struct in the general case (as apposed to the tag of the specific type) that can be used for a workaround? Thanks in advance, Mike #include <boost/fusion/adapted/struct/adapt_struct.hpp> #include <boost/fusion/container/vector/vector.hpp> #include <boost/fusion/algorithm/query/find.hpp> #include <boost/type_traits.hpp> #include <string> #include <iostream> namespace b = boost; namespace bf = boost::fusion; namespace demo { struct employee { std::string name; int age; }; } // demo::employee is now a Fusion sequence BOOST_FUSION_ADAPT_STRUCT( demo::employee, (std::string, name) (int, age)) int main(void) { typedef bf::vector<int,float> V1; // Find against non-const V1 V1 v1(1,2.0); BOOST_MPL_ASSERT_NOT(( b::is_same<bf::result_of::find<V1,int>::type,bf::result_of::end<V1>::type> )); assert(*bf::find<int>(v1) == 1); // Find against const V1 const V1 const_v1(3,4.0); BOOST_MPL_ASSERT_NOT(( b::is_same<bf::result_of::find<const V1,int >::type,bf::result_of::end<const V1>::type> )); assert(*bf::find<int>(const_v1) == 3); // Find against non-const employee demo::employee e1 = {"Foo",18}; BOOST_MPL_ASSERT_NOT(( b::is_same<bf::result_of::find<demo::employee,int>::type,bf::result_of::end<demo::employee>::type> )); assert(*bf::find<int>(e1) == 18); // Find against const employee const demo::employee const_e1 = {"Bar",21}; #if 0 BOOST_MPL_ASSERT_NOT(( b::is_same<bf::result_of::find<constdemo::employee, int>::type,bf::result_of::end<const demo::employee>::type> )); assert(*bf::find<int>(const_e1) == 21); #else // note that the find key is a const int for this to work and is inconsistent with fusion::vector BOOST_MPL_ASSERT_NOT(( b::is_same<bf::result_of::find<constdemo::employee, const int>::type,bf::result_of::end<const demo::employee>::type> )); assert(*bf::find<const int>(const_e1) == 21); #endif return 0; }

2 2

[mpi] irecv / send problem
by Nick Collier 06 Aug '09

06 Aug '09

I running into an issue where an irecv followed by a send results in deadlock. A simple test case, class Item { private: friend class boost::serialization::access; template<class Archive> void serialize(Archive& ar, const unsigned int version) { ar & val; } public: int val; Item() : val(1) { } }; struct Receipt { boost::mpi::request request; std::vector<Item> items; }; int main(int argc, char **argv) { mpi::environment env(argc, argv); mpi::communicator world; Receipt receipt; vector<Item> msg(100000); int other = world.rank() == 0 ? 1 : 0; cout << world.rank() << " irecv from " << other << endl; receipt.request = world.irecv(other, 0, receipt.items); cout << world.rank() << " sending to " << other << endl; world.send(other, 0, msg); receipt.request.wait(); cout << "Done" << endl; } Run with mpirun -np 2, this never completes. It does complete with vector<Item> msg(10) however. Nick

5 7

[lambda] Help figuring out variable substitution, lazy functions, etc.? Should I be looking at Phoenix
by Jesse Perla 06 Aug '09

06 Aug '09

I have been trying to prevent variable substitution with boost::lambda and cannot figure out how to prevent variable substitution. Here is a (simplified) version: double first_moment(double x) { return x; } //And create a function auto myfunc = first_moment(_1); cout << myfunc(2.0); ... This doesn't compile, and every permutation I can think of using protect, etc. fails as well. Now I realize that this looks like a job for "bind" (which works fine of course) but this is a simplified pattern of the notation I want. Is there any way to do this with protect, unlambda, etc. around the _1? Are other libraries like Phoenix a better place to look for this kind of functionality? Do people see Phoenix as the evolution of boost lambda or are these likely to remain in parallel? Should I stick with C++0X lambdas? Thanks, Jesse

3 3

Terminate an io_service properly
by Robert Gruendler 06 Aug '09

06 Aug '09

Hi, i'm trying to create a simple socket server wich should be able to shut down when receiving a specific message. I've tried to alter the blocking tcp echo server example of the boost::asio library, but i'm not sure how to properly end the main "listening" loop from within the session thread. here's the modified asio example. when i try to close the acceptor from within the session thread, i get an exception "*The file handle supplied is not valid". *i assume this is the wrong place to close the acceptor / stop the io_service, but i don't know where else i can do this: using boost::asio::ip::tcp; const int max_length = 1024; typedef boost::shared_ptr<tcp::socket> socket_ptr; boost::asio::io_service io_service; tcp::acceptor* acceptor; void session(socket_ptr sock) { try { for (;;) { char data[max_length]; boost::system::error_code error; size_t length = sock->read_some(boost::asio::buffer(data), error); if (error == boost::asio::error::eof) break; // Connection closed cleanly by peer. else if (error) throw boost::system::system_error(error); // Some other error. std::string message = data; if (message.compare("SHUTDOWN") == 0) { //how to terminate the server through an incoming message ? acceptor->close(); io_service.stop(); } boost::asio::write(*sock, boost::asio::buffer(data, length)); } } catch (std::exception& e) { std::cerr << "Exception in thread: " << e.what() << "\n"; } } void server(boost::asio::io_service& io_service, short port) { acceptor = new tcp::acceptor(io_service, tcp::endpoint(tcp::v4(), port)); std::cout << "server started" << std::endl; for (;;) { if (acceptor->is_open()) { socket_ptr sock(new tcp::socket(io_service)); acceptor->accept(*sock); boost::thread t(boost::bind(session, sock)); } else break; } std::cout << "server stopped"; } int main(int argc, char* argv[]) { try { server(io_service, 22222); } catch (std::exception& e) { std::cerr << "Exception: " << e.what() << "\n"; } delete acceptor; return 0; }

3 2

Boost Software License implications for derivative works
by Naik, Roshan 06 Aug '09

06 Aug '09

The Boost Software License (BSL) requires derivative works to include the text of the Boost License itself. I am considering the use of the Boost preprocessor library (only) to implment certain parts of Castor <http://www.mpprogramming.com/cpp> (an open source C++ library) which is already covered under the MIT license. My real concern is, by including the BSL text in Castor, I dont want it to look as if Castor is covered by BSL when it is actually covered by MIT license. As Boost::pp would be used internally it would be shipped as part of the library. Would it suffice if boost::pp headers were simply bundled as-is, along with the licensing text that appears at the top each header ? That way it would appear as if only those headers are covered by BSL and the rest of Castor is covered by MIT. - Roshan

3 2

Why does auto_link.hpp assume that debug versions of STLPort are built with _STLP_DEBUG?
by Max Motovilov 06 Aug '09

06 Aug '09

Subj. Is there a [relatively non-invasive] way to break this dependency? As far as I can tell, STLPort itself makes no such assumption and builds in non-_STLP_DEBUG mode by default. I cannot use _STLP_DEBUG mode for other reasons... The platform is Windows Mobile 5, Visual Studio 2008 (VC9) Regards, ...Max...

2 3