- Boost-users - lists.boost.org

bug in asio?
by Hei Chan 23 Aug '10

23 Aug '10

Hi, I am using a 3rd party library that uses boost::asio. The 3rd party library will call asyn_receive() with a callback handler, then when boost::asio::detail:: receive_op::do_complete() calls the callback handler, the handler consumes the message, calls asyn_receive() and then if there is no more message, it (the thread) waits for a condition -- the callback handler gets called again. The problem now is that once there is no more message, the thread enters the wait condition, and it never gets waken up again. Is it incorrect to enter the wait condition in the same thread that calls boost::asio::detail:: receive_op::do_complete()? Thanks in advance. Cheers, Hei

1 0

Getting started with Boost.Test
by Otto 23 Aug '10

23 Aug '10

Hi, I'm very new to C++ and Boost. I want to make a simple desktop application (a game editor, basically read/write a binary file) using QtCreator. Right now I'm only interested in Boost.Test for the unit tests. I tried to install using the instructions in "Getting started" but wasn't able to, probably because of bad english and lack of C++ knowledge. I unpacked Boost to /home/otto/boost_1_44_0/ >From there, I ran: $ mkdir dist $ ./bootstrap.sh --prefix=/home/otto/boost_1_44_0/dist --show-libraries --with-libraries=test $ ./bjam install It seems that config had no effect as it built everything (instead of just test) and to bin.v2/. A few "cp: cannot create regular file `/usr/local/lib/libboost_wave.a': Permission denied" errors too. What went wrong? Now I have bin.v2/libs/test/build/gcc-4.4.4/release with directories link-static and threading-multi. I also would like to configure it for usage with QtCreator. Thanks

3 5

Best method for making synchronous call from async handler
by David Daeschler 23 Aug '10

23 Aug '10

Good evening, Currently I have an application running on boost asio working very well. To simplify the heck out of it, it takes requests via tcp and passes back binary information stored on disk. However, I now need to add an extra step to processing. If the data is not available on the local disk, the server needs to contact one or more peer servers it knows about, send a query to each one via UDP asking if it has the data, and then use an existing TCP connection to that server to retrieve the asset. This step must be completed before any more processing can be done for the current client even though the client may be sending even more requests while we process. My question is, what is the best method for me to achieve the goal? Should I set up another io_service for the UDP query step along with retrieving the data from the other server? Is there a good wait to wait on that io_service to complete? Also in the case that one of the servers has gone away, can I use a deadline timer to cancel the UDP recvfrom if I don't get a response in time from one or more servers? I should also probably keep this second io_service alive throughout the lifetime of the application? Thank you in advance for advice on tackling this problem the right way with asio. -- David Daeschler

2 2

mpl::fold problems
by sergey kostanbaev 23 Aug '10

23 Aug '10

Hello everybody! I'm a newbie in boost and I've got some problems with boost::mpl #include <boost/mpl/vector.hpp> #include <boost/mpl/fold.hpp> #include <boost/mpl/arithmetic.hpp> template <class T, unsigned space=1> struct tr_field2 { static const unsigned count = space; } typedef boost::mpl::vector4< tr_field2<int, 4>, tr_field2<int, 7>, tr_field2<char, 1>, tr_field2<char, 1> > Test_description; I want to calculate a sum of all tr_field2::count I've read in the docs mpl::fold is for the case. I tried to compose typedef boost::mpl::fold<Test_description, boost::mpl::int_<0>, boost::mpl::plus<boost::mpl::_, boost::mpl::_::count > > >::type fcount; but it won't compile. It seems I misundestood something.

2 2

[BGL] Change order of edge iterator
by John Robertson 23 Aug '10

23 Aug '10

I want to be able to iterate through the edges in any given order, not just the order specified by tie(edge_iter, edges_end) = edges(g);. Is it possible to do something like the following. typename graph_traits<Graph>::edge_iterator edge_iter1, edges_end1, edge_iter2, edges_end2, edge_iter3, edges_end3; tie(edge_iter1, edges_end1) = edges(g); sort(edge_iter1, edges_end1, ordering_1()); tie(edge_iter2, edges_end2) = edges(g); sort(edge_iter2, edges_end2, ordering_2()); tie(edge_iter3, edges_end3) = edges(g); sort(edge_iter3, edges_end3, ordering_3()); (ordering_x() specified by some struct) Then I would have 3 different orderings of the edges available. Would this work or should I use an iterator adapter, or create a map and (if its possible) obtain an iterator from the mapping, or something else. Thanks

2 1

[DateTime] from time_t to ptime and back - or not
by Christoph Duelli 23 Aug '10

23 Aug '10

[I am using Boost 1.43.0 on linux, gcc 4.4.1] I am having trouble with date_time's ptime. I am passing it a time_t, and store that in a ptime object. When I retrieve the time_t, I am getting a different value than the one I passed to it originally. (the difference seems to be 2 hours - in all probability due to UTC vs local time issues). Code: time_t now = time(NULL); boost::posix_time::ptime pt = boost::posix_time::from_time_t(now); struct tm tm = to_tm(pt); time_t should_be_now = mktime(&tm); BOOST_CHECK_EQUAL(now, should_be_now); fails for me: now is by 7200 seconds (i.e. 2 hours) ahead of. Is this a bug in date_time or am I just failing to use it properly? Thank you, and best regards Christoph

3 3

[bind]Help! Can we modify parameters before pass them into bind (see the context below for details)
by jungleman 23 Aug '10

23 Aug '10

Hi, I have some problems on function template bind. the original usage of bind is : bind(func, _1, _2), and I wonder whether I can modify the parameters before pass them into bind, for example, functype function = bind(func, *_1, *_2). So that, when I use function(a, b), it will do like this: func(*a, *b). May it work, or some way else to do this? I just wanna do this in one line code. thanks in advance. -- Stay hungry, stay foolish !

3 2

Boost.Intrusive issues on IBM compiler
by Jeremiah Willcock 22 Aug '10

22 Aug '10

It appears that almost all of the Boost.Intrusive tests (at <URL:http://tinyurl.com/2bzhglh>, for example) fail on IBM's compiler. I get similar results, like the following from the test suite, from my own code: An expression of type "overloaded function: U::to_node_ptr" cannot be converted to type "boost::intrusive::list_node<void *> *(boost::intrusive::detail::base_hook_traits<boost::intrusive::testvalue<hooks<void *>,0>,boost::intrusive::list_node_traits<void *>,1,boost::intrusive::default_tag,1>::*)(testvalue<hooks<void *>,0> &)" Is there an easy workaround for this issue? Is there a plan to fix it? This has also been reported a couple of months ago as bug 4412 (<URL:https://svn.boost.org/trac/boost/ticket/4412>). Thank you. -- Jeremiah Willcock

2 1

axpy_prod Weirdness
by Jane Shevtsov 22 Aug '10

22 Aug '10

I have a little program that implements an ecological network analysis algorithm. (It looks at the importance of indirect energy or matter flows between species.) 01 //dea.cpp -- The fundamental DEA algorithm 02 03 #include <boost/numeric/ublas/matrix.hpp> 04 #include <boost/numeric/ublas/operation.hpp> 05 #include <boost/numeric/ublas/io.hpp> //for test 06 #include "storage_adaptors.hpp" //for test 07 08 using boost::numeric::ublas::matrix; 09 10 void dea(const matrix<double>[], int, int, matrix<double>[]); 11 matrix<double> indflow(const matrix<double> & Ndea, const matrix<double> & Gdea); 12 13 int main() { 14 using boost::numeric::ublas::make_matrix_from_pointer; 15 matrix<double> Garr[4]; 16 matrix<double> Narr[2]; 17 double G1[3][3] = {0, 0, 0, 0.33333, 0, 0, 0.66667, 1, 0}; 18 double G2[3][3] = {0, 0, 0, 0.6, 0, 0, 0.4, 1, 0}; 19 double G3[3][3] = {0, 0, 0, 0.55, 0, 0, 0.45, 1, 0}; 20 double G4[3][3] = {0, 0, 0, 0.5, 0, 0, 0.5, 1, 0}; 21 Garr[0] = make_matrix_from_pointer(G1); 22 Garr[1] = make_matrix_from_pointer(G2); 23 Garr[2] = make_matrix_from_pointer(G3); 24 Garr[3] = make_matrix_from_pointer(G4); 25 dea(Garr, 2, 4, Narr); 26 std::cout << Narr[0] << std::endl; 27 matrix<double> ratio = indflow(Narr[0], Garr[0]); 28 std::cout << ratio << std::endl; 29 return 0; 30 } 31 32 //The fundamental DEA algorithm 33 //Takes array of G matrices, window size, array length and empty array for N_DEA matrices as input 34 void dea(const matrix<double> Garr[], int win, int arrLen, matrix<double> Narr[]) { 35 using namespace boost::numeric::ublas; 36 int matSize = Garr[0].size1(); 37 for (int i = 0; i < arrLen - win; i++) { 38 matrix<double> Gprod = identity_matrix<double>(matSize); 39 Narr[i] = identity_matrix<double>(matSize); 40 //the core loop 41 for (int j = 0; j < win; j++) { 42 Gprod = prod(Gprod, Garr[i+j]); 43 Narr[i] += Gprod; 44 } 45 } 46 } 47 48 //computes indirect flow fraction 49 matrix<double> indflow(const matrix<double> & Ndea, const matrix<double> & Gdea) { 50 using namespace boost::numeric::ublas; 51 int matSize = Ndea.size1(); 52 matrix<double> iflow = Ndea - Gdea - identity_matrix<double>(matSize); //Ind = N-G-I 53 matrix<double> frac = zero_matrix<double>(matSize, matSize); //inialize indirect flow fraction matrix 54 //perform division for nonzero, off-diagonal elements 55 for (int i=0; i<matSize; i++) { 56 for (int j=0; j<matSize; j++) { 57 if (i != j && Ndea(i,j) != 0) //avoid division by zero 58 frac(i, j) = iflow(i, j) / Ndea(i, j); 59 } 60 } 61 return frac; 62 } The correct output is below. [3,3]((1,0,0),(0.33333,1,0),(1.26667,1,1)) [3,3]((0,0,0),(0,0,0),(0.473683,0,0)) The code given above works, but to speed it up, I tried replacing the first line of the core loop in dea() with axpy_prod(Gprod, Garr[i+j], Gprod, true);. This gives the following wrong output: [3,3]((1,0,0),(0,1,0),(0,0,1)) [3,3]((0,0,0),(0,0,0),(0,0,0)) And if I use axpy_prod(Gprod, Garr[i+j], Gprod, false);, I get: [3,3]((3,0,0),(1.26666,3,0),(2.33334,3,3)) [3,3]((0,0,0),(0.736843,0,0),(0.714285,0.666667,0)) What gives? When should I avoid using axpy_prod? -- ------------- Jane Shevtsov Ecology Ph.D. candidate, University of Georgia co-founder, <www.worldbeyondborders.org> Check out my blog, <http://perceivingwholes.blogspot.com>Perceiving Wholes "The whole person must have both the humility to nurture the Earth and the pride to go to Mars." --Wyn Wachhorst, The Dream of Spaceflight

1 0

[bind]Help! Can we modify parameters before pass them into bind (see the context below for details)
by jungleman 22 Aug '10

22 Aug '10

Hi, I have some problems on function template bind. when I uses stl algorithm adjacent_find(inputIterator begin, inputIterator) -- Stay hungry, stay foolish !

2 1