- Boost-users - lists.boost.org

array alignment in ublas matrices
by petros 13 Sep '11

13 Sep '11

Hi, I am using the ublas matrices with mkl. The documentation of mkl suggests/requires that allocated arrays are aligned on 16 byte boundaries. The matrix structures within ublas delegate the memory management in the std allocator which at the very end reduces to new. I know that one way to address this is write my own allocator for the unbounded_array that lies on the basis of the storage structures. However, this is an error-prone job and it is suggested that this is avoided. Any other solutions? – hopefully more straightforward ? Thank you in advance for your help, Petros ps: using visual studio 2010 on winn7/visa 64 bit and mkl 10.3.

2 2

[serialization] yaml archive compilation
by alfC 12 Sep '11

12 Sep '11

Hi, did any body succeed in compiling the YAML component of Boost.Serialization located in https://github.com/boost-vault/serialization ? I really don't understand what should I do with the yaml_oarchive.cpp yaml_iarchive.cpp, etc, contained in the src directory. The source don't have makefiles, should I put those in the boost source tree? Thanks, Alfredo

1 0

LINK Error while building PBGL example breadth_first_search.cpp
by 余韬 12 Sep '11

12 Sep '11

Hi, I am trying to learn PBGL. I am using VS2008 on Win 7. I had built and installed open mpi 1.4.3, HPC Pack 2008 SDK, and Boost library 1.46.1 with "bjam --with-mpi --with-graph_parallel". I had set include and library directory for the above 3 libraries in VS2008. The additional dependency set for the project is "libboost_*mpi*-vc90-mt-gd.lib libboost_*serialization*-vc90-mt-gd.lib libboost_wserialization-vc90-mt-gd.lib libboost_graph-vc90-mt-gd.lib libboost_*graph_parallel*-vc90-mt-gd.lib libboost_system-vc90-mt-gd.lib libmpi_cxxd.lib libmpid.lib libopen-pald.lib libopen-rted.lib msmpi.lib" Here are the major references I had referred which helped to reduce errors significantly: http://terboven.wordpress.com/2008/09/09/building-and-using-boostmpi-on-win… http://lists.boost.org/boost-users/2009/11/54042.php However, while compiling D:\Development\boost_1_46_1\libs\graph_parallel\example\breadth_first_search.cpp on my machine, I still got the following 4 link errors: Error 4 error LNK2019: unresolved external symbol "*private: struct ompi_datatype_t * __thiscall boost::mpi::detail::mpi_datatype_map::get(class type_info const *)*" breadth_first_search.obj Error 3 error LNK2019: unresolved external symbol "*private: void __thiscall boost::mpi::detail::mpi_datatype_map::set(class type_info const *,struct ompi_datatype_t *)*" breadth_first_search.obj Error 2 error LNK2019: unresolved external symbol "*public: __thiscall boost::mpi::communicator::operator struct ompi_communicator_t *(void)const*" breadth_first_search.obj Error 1 error LNK2019: unresolved external symbol "void __cdecl boost::detail::bfs_helper<class boost::adjacency_list<struct boost::vecS,struct boost::distributedS<class boost::graph::distributed::mpi_process_group,struct boost::vecS,struct boost::defaultS>......" breadth_first_search.obj Am I missing something here? Thanks, Tao Yu http://about.me/Tao_Yu

1 1

graph documentation dead link
by Benjamin Sobotta 12 Sep '11

12 Sep '11

Hello, I'm currently reading about the graph library, more specifically about the shortest path problem with resource constraints, and found that the link to the paper Irnich, S.; Desaulniers, G. (2005): Shortest Path Problems with Resource Constraints in: Desaulniers, G.; Desrosiers, J.; Solomon, M. (eds.) (2005): Column Generation Springer, New York, pp. 33–65 is broken. I found a new link to the paper: http://logistik.bwl.uni-mainz.de/Dateien/or_2004-01.pdf Cheers, Ben PS: The link is in .../doc/libs/1_47_0/libs/graph/doc/r_c_shortest_paths.html under "Introduction and Problem Description"

2 1

thread running many times
by Kraus Philipp 12 Sep '11

12 Sep '11

Hello, I create a thread group and run the group with join_all(). My code shows like: for(i=0; i < runs; ++i) { threadgoup.join_all(); do something with the results } So I run the thread group n times. Is this the correct way or can I call join_all only once on the group? Thanks PHil

3 8

[asio] serial_port::async_read_some() with native COM ports on WIN32
by Running Razor 12 Sep '11

12 Sep '11

I use serial_port::async_read_some() to read data from serial interface on windows. Using virtual com ports (FTDI, SiLabs) everything wents as expected. But on native COM ports the completion handler is not called until the buffer (passed to async_read_some) is completely filled. I use "plain" communication without hardware handshake: ... typedef serial_port_base ser; mDevice.set_option( ser::baud_rate ( baud ) ); mDevice.set_option( ser::flow_control ( ser::flow_control::none ) ); mDevice.set_option( ser::parity ( ser::parity::none ) ); mDevice.set_option( ser::stop_bits ( ser::stop_bits::one ) ); mDevice.set_option( ser::character_size( 8 ) ); ... Do I miss something or is this a bug? tested on boost_1_38_0 and boost_1_46_1

1 0

boost thread (non-blocking) communication
by Jack Bryan 11 Sep '11

11 Sep '11

Hi, I am trying to do data-exchange communication among boost::threads. Are there asynchronous (non-blocking) communication APIs in boost ? For example, non_block_send( receiver_address, data_buffer, data_volume, data_tag , send_status ); non_block_recv( sender_address, data_buffer, data_volume, data_tag , recv_status ); if (recv_status == TRUE) getData(data_buffer) ; else wait; Any help is really appreciated. Thanks

4 4

Compile errors using boost 4.7.4 and Intel ICC x64 on Debian
by Bill Moo 11 Sep '11

11 Sep '11

Hello List. The attached file contains the errors I get when I attempt to compile my utilities library using Boost 1.47.0 with the Intel Compiler v12.1.0 on Debian Squeeze v6.0.2. (My IDE is Qt Creator v2.3.0 using v4.7.4 (x64) SDK) This is my first attempt at doing this under Linux as the utilities library (dll) compile as is in Windows. So I would appreciate your thoughts on this as I can't continue the migration of the other projects until I get this library working. Thank you. -- Bill

1 0

dijkstra_shortest_paths between sets of nodes (between polygons on a grid)
by Stuck Developer 10 Sep '11

10 Sep '11

This ought to be a classic problem but I haven't found a solution yet... Conceputally, I'm trying to find the shortest paths between all polygons within a maximum weighted distance over a grid of friction weights (as well as the total weighted distance of each of these shortest paths). The grid with all the weights (a "friction" value representing the weight to move over that particular grid cell) is currently used to create a network with one vertex per grid cell and edges to its immediate eight neighbours (in the cardinal and diagonal directions). The weight of each edge is calculated to the average weight of the vertex and its neigbours in the four cardinal directions and multiplied by sqrt(2) in the diagonal directions to compensate for the longer euclidean distance. Conceptually again, on this grid, there are a bunch of polygons. These are then made up ofcontiguous groups of vertices. Each polygon could thus be considered a set (or a component) of vertices with a unique id for each polygon. The problem now is to find the shortest path over the weighted grid between the edges of all polygons within a maximum weighted distance from each other. NOTE that the shortest path between two polygons should be able to pass through the weighted grid cells of other polygons along the way when these other polygons are not considered being source or target "vertex sets". I first tried to solve this by looping through all of the polygons and finding the shortest path between some representative vertex (e.g. the centroid grid cell) inside a source polygon to a representative vertex within each possible target polygon. I did this by setting all of the grid cell weights to zero within each of the two polygons, running the dijkstra_shortest_paths between their representative vertices and then resetting the weights within these two polygons before solving it for another pair of polygons. This actually works but the search goes bezerk within the polygons where all the edges have a cost of zero, and if these polygons are big enough, I run out of memory (I guess the shortest path becomes too long...). I'm thinking this should be possible to solve by instead initializing all of the vertices within each source polygon to "discovered" (i.e. black) somehow and initializing the priority queue and the cost in some corresponding manner. I just don't know how. Then there is the question of knowing when the shortest path to the target polygon has been reached... And thirdly if it would be possible to stop the search when a maximum weighted distance has been reached. Any smart solutions to this problem?

2 1

Re: [Boost-users] [bind] Dereferencing placeholder argument
by Ivan Le Lann 09 Sep '11

09 Sep '11

----- "Szymon Gatner" <szymon.gatner(a)gmail.com> a écrit : > 2011/9/9 Nat Linden <nat(a)lindenlab.com>: > > On Fri, Sep 9, 2011 at 10:04 AM, Ivan Le Lann <ivan.lelann(a)free.fr> > wrote: > > >> Have you tried *_1 ? > > > > I think that works with boost::lambda::bind and boost::lambda::_1 > > (also, presumably, the corresponding constructs in Phoenix) but not > > with plain boost::bind. > > Indeed I tried that - getting invalid indirection error. Some time ago I offered a patch to add arithmetical operator support to Boost.Bind and was answered : Use Lambda or Phoenix. If you want to tweak Boost.Bind, which I would not encourage, maybe this could help : http://boost.2283326.n4.nabble.com/Bind-Function-arithmetic-sample-impl-ins… As said, it is for arithmetical operators but might be adapted to your need. Regards, Ivan

1 0