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

is it possible to group values with Boost's rtree?
by Andy Edwards 27 Aug '15

27 Aug '15

Hi everyone, I need a way to take a collection of 3D objects and put them into disjoint minimally overlapping groups no bigger than some N. Can this be done with Boost's rtree or any other Boost libraries? I've done this by putting the objects in an R* tree and using the nodes at a certain level of the tree (for example, 2 levels up from the leaf level) to determine the groups. However, that was in my own Java R* tree implementation that provided read-only access to its internal nodes. I'd like to port this program to C++ using Boost's rtree, but it doesn't provide read-only access to its nodes, does it? Does anyone know if it would be possible to extend the rtree to provide this? Or would it be possible to write a custom predicate to get the groupings? Thanks! Andy

3 2

[filesystem] Relative Draft Proposal RFC
by Beman Dawes 27 Aug '15

27 Aug '15

There have been requests for a Filesystem library relative function for at least ten years. The requested functionality seems simple - given two paths with a common prefix, return the non-common suffix portion of one of the paths such that it is relative to the other path. In terms of the Filesystem library, path p("/a/b/c"); path base("/a/b"); path rel = relative(p, base); // the requested function cout << rel << endl; // outputs "c" assert(absolute(rel, base) == p); If that was all there was to it, the Filesystem library would have had a relative function years ago. Critical issues: Clashing requirements, symlinks, directory placeholders (dot, dot-dot), user-expectations, corner cases. A paper by Jamie Allsop, Additions to Filesystem supporting Relative Paths, is what broke my mental logjam. Much of [my proposal] is based directly on Jamie's analysis and proposal. The weakly_canonical function and aspects of the semantic specifications are my contributions. Mistakes, of course, are mine. The full proposal is at http://boostorg.github.io/filesystem/relative_proposal.html A preliminary implementation is available in the feature/relative2 branch of the Boost Filesystem Git repository. See github.com/boostorg/filesystem/tree/feature/relative2 The overall plan is to get feedback from boost developers, then merge to develop for boost 1.60, then propose to the C++ committee for addition to the Filesystem TS. Comments welcome. If you are one of the many people who have requested "relative" functionality, does this proposal meet your needs? If not, why? Thanks, --Beman

5 7

Avoid including path to BOOST into exe
by Pavel Seleznev 27 Aug '15

27 Aug '15

Hello, I have the same question as described here http://stackoverflow.com/questions/32141335/avoid-including-path-to-boost-i… I have built BOOST with the following command line: b2 --toolset=msvc-10.0 --stagedir=./stage/2010/win32 address-model=32 variant=release threading=multi link=static runtime-link=static then compiled a project with statically linked runtime and release configuration with NDEBUG, BOOST_DISABLE_ASSERTS defined. Compiler options: c++\advanced\use full paths (/FC) is off c++\general\debug information format (/ZI /Zi /Z7) is empty ...and still having the entries in my binary file: c:\work\boost/xpressive/basic_regex.hpp void __cdecl boost::xpressive::detail::throw_on_stack_error(bool) Regex stack space exhausted c:\work\boost/xpressive/detail/dynamic/parse_charset.hpp void __thiscall boost::xpressive::detail::char_overflow_handler::operator () How can I avoid this? Regards, Pavel

1 0

Using boost modified bessel functions with complex arguments
by Ross Kynch 27 Aug '15

27 Aug '15

Dear all, Does anyone know how to call boost's modified bessel function of the first kind using a complex argument? Specifically I want the positive half integer order with complex argument so I use: #include <boost/math/special_functions/bessel.hpp> std::complex<double> output = boost::math::cyl_bessel_i(double nhalf, std::complex<double> x); but this gives me a huge list of compiler errors which seems to suggest the std::complex<double> template isn't valid. I am able to call other boost math library functions so I don't believe my install is to blame here. Thanks for any help Ross

2 2

Asio io_service and Thread scheduler/executor
by Mikael Olenfalk 26 Aug '15

26 Aug '15

Hi, I am currently thinking about how to move our project at work from a set of specific threads (1 thread per module e.g.) and our own network library abstraction to Boost.Asio and Boost.Thread's scheduler/executors. Before I read about schedulers/executors I considered just using asio::io_service (with a pool of threads calling .run()) for quick-shot background offloading, but after looking at scheduler/executor I would prefer using it for background work. Does anybody have any experience how to combine boost::asio with boost::thread's scheduler/executors? In my preferred design I would only have one thread pool backing both asio work and background computational tasks (e.g. using boost::async(Executor&, F&&, ...) ) My initial gut feeling was to have a thread-pool calling boost::asio::io_service.run() and then wrap that in an executor. I guess what I am asking for if somebody has any previous experience to share with this or similar approaches. Kind regards, /M

2 1

Fw: Use of QMutex not as expected
by Don El Ron 26 Aug '15

26 Aug '15

3 2

Boost Social Event at CppCon
by Jon Kalb 25 Aug '15

25 Aug '15

Boosters, Because a lot of Boosters are going to be attending CppCon, we¹d like to arrange a social event for Boosters/C++Now regulars. http://cppcon.org/ If you are planning to attend CppCon or even If you are just going to be in or around Bellevue, Washington between September 20th and 25 and you¹d like to join us for dinner, please let us know so we can get a count. Please send mail to: social(a)boost.org Once we settle on a date, etc, we¹ll let everyone who responded know when and where, so please let us know, even if you are a ³maybe.² I hope to see you at CppCon. Jon

1 0

[Graph] Distributed edge insertion
by Sensei 25 Aug '15

25 Aug '15

Dear all, I am finding difficulties in understanding how distributed graphs work. My graph is simple, a string-valued property, with some edges, however, right now I have only one edge... if it works. Since each process will add its own nodes, is it possible to avoid communication when adding nodes, given that each node owns its own vertices, and no overlap is possible between processes? This would save some communication avoiding re-distribution of nodes. Can I add edges concurrently? Meaning, each process adds some edges. I've only seen examples with root adding edges, but that's limiting. My last question is maybe easy. I'm adding an edge that connects two nodes, owned by different processes. Doing this results in a crash. It seems I cannot access properties with operator[] when traversing processes. I'm posting the code below, along with the *crashing* output. I hope you can help getting a grasp on PBGL! Thanks! #include <iostream> #include <string> #include <vector> #include <cmath> #include <list> #include <functional> #include <boost/config.hpp> #include <boost/mpi.hpp> #include <boost/graph/use_mpi.hpp> #include <boost/graph/distributed/mpi_process_group.hpp> #include <boost/graph/distributed/adjacency_list.hpp> #include <boost/graph/iteration_macros.hpp> using namespace boost; using boost::graph::distributed::mpi_process_group; // Useless class class str { public: str(std::string s = std::string()) : s_(s) { // NOP } template<typename Archiver> void serialize(Archiver& ar, const unsigned int /*version*/) { ar & s_; } std::string s_; }; // Allow indexing and constructing namespace boost { namespace graph { // Index based on the str class template<> struct internal_vertex_name<str> { typedef multi_index::member<str, std::string, &str::s_> type; }; // Construct with str class template<> struct internal_vertex_constructor<str> { typedef vertex_from_name<str> type; }; } } // Handy typedef boost::adjacency_list<vecS, distributedS<mpi_process_group, vecS>, bidirectionalS, str> dgraph; typedef graph_traits<dgraph>::vertex_descriptor dvertex; // START ME UP int main(int argc, const char * argv[]) { boost::mpi::environment env; boost::mpi::communicator comm; std::cout << "RANK " << comm.rank() << std::endl; comm.barrier(); dgraph g; std::vector<dvertex> descriptors; // Each rank for(int i = 0; i < 10; i++) { str s(std::to_string(i + (comm.rank() + 1) * 100)); std::cout << "rank " << comm.rank() << " adding " << s.s_ << std::endl; descriptors.push_back( add_vertex(s, g) ); } BGL_FORALL_VERTICES(v, g, dgraph) { std::cout << "V @ " << comm.rank() << " " << g[v].s_ << " local " << v.local << " owner " << v.owner << std::endl; } std::cout << ">>> NUM VERTICES " << num_vertices(g) << " NUM EDGES " << num_edges(g) << std::endl; comm.barrier(); if(comm.rank() >= 0) { int from, dest; bool letsCrash = true; if (letsCrash) { from = 1; dest = 0; } else { from = 0; dest = 0; } std::cout << "rank " << comm.rank() << " adding edge owners " << descriptors[from].owner << " - " << descriptors[dest].owner << std::endl; add_edge(descriptors[from], descriptors[dest], g); } boost::synchronize(g); BGL_FORALL_EDGES(e, g, dgraph) { std::cout << "E @ " << comm.rank() << " " << g[boost::source(e, g)].s_ << " -> " << g[boost::target(e, g)].s_ << std::endl; } return 0; } RANK 0 RANK 1 rank 0 adding 100 rank 0 adding 101 rank 1 adding 200 rank 1 adding 201 rank 0 adding 102 rank 0 adding 103 rank 1 adding 202 rank 1 adding 203 rank 0 adding 104 rank 1 adding 204 rank 0 adding 105 rank 1 adding 205 rank 1 adding 206 rank 0 adding 106 rank 0 adding 107 rank 1 adding 207 rank 0 adding 108 rank 1 adding 208 rank 0 adding 109 rank 1 adding 209 V @ 0 100 local 0 owner 0 V @ 0 201 local 1 owner 0 V @ 0 102 local 2 owner 0 V @ 0 203 local 3 owner 0 V @ 0 104 local 4 owner 0 V @ 0 205 local 5 owner 0 V @ 1 200 local 0 owner 1 V @ 1 101 local 1 owner 1 V @ 1 202 local 2 owner 1 V @ 1 103 local 3 owner 1 V @ 1 204 local 4 owner 1 V @ 1 105 local 5 owner 1 V @ 1 206 local 6 owner 1 V @ 1 107 local 7 owner 1 V @ 1 208 local 8 owner 1 V @ 0 106 local 6 owner 0 V @ 0 207 local 7 owner 0 V @ 0 108 local 8 owner 0 V @ 0 209 local 9 owner 0 >>> NUM VERTICES 10 NUM EDGES 0 V @ 1 109 local 9 owner 1 >>> NUM VERTICES 10 NUM EDGES 0 rank 1 adding edge owners 0 - 1 rank 0 adding edge owners 1 - 0 Assertion failed: (v.owner == processor()), function operator[], file /usr/local/include/boost/graph/distributed/adjacency_list.hpp, line 1696. Assertion failed: (v.owner == processor()), function operator[], file /usr/local/include/boost/graph/distributed/adjacency_list.hpp, line 1696. E @ 0 201 -> E @ 1 101 -> [Senseis-MBP:40908] *** Process received signal *** [Senseis-MBP:40908] Signal: Abort trap: 6 (6)

1 1

boost::ptr_vector does not support const types
by Martin Gamwell Dawids 25 Aug '15

25 Aug '15

I would like to use a ptr_vector to hold a number of const values but it seems ptr_vector does not currently support the use of const value types. See this simple code example: #include "boost/ptr_container/ptr_vector.hpp" #include <vector> class X { const int i; public: X(int i) : i(i) {} int getI() const { return i; } }; int main() { boost::ptr_vector<X> bvx; boost::ptr_vector<const X> bvcx; std::vector<X*> svx; std::vector<const X*> svcx; bvx.push_back(new X(42)); bvcx.push_back(new X(42)); // does not compile svx.push_back(new X(42)); svcx.push_back(new X(42)); return 0; } I have tried this on MSVC 2005/8.0 and with g++ on Mac – i.e. Apple LLVM version 6.1.0 (clang-602.0.53) (based on LLVM 3.6.0svn). On both compilers, calling 'push_back()' on a 'boost::ptr_vector<const X>' does not compile and on g++/Mac it yields this error message: ../boost/boost/ptr_container/ptr_sequence_adapter.hpp:249:37: error: reference to type 'const value_type' (aka 'void *const') could not bind to an lvalue of type 'value_type' (aka 'const X *') this->base().push_back( x ); // strong, commit ^ ptr_vector_test.cpp:22:8: note: in instantiation of member function 'boost::ptr_sequence_adapter<const X, std::__1::vector<void *, std::__1::allocator<void *> >, boost::heap_clone_allocator>::push_back' requested here bvcx.push_back(x); ^ /Library/Developer/CommandLineTools/usr/bin/../include/c++/v1/vector:697:62: note: passing argument to parameter '__x' here _LIBCPP_INLINE_VISIBILITY void push_back(const_reference __x); ^ 1 error generated. So the problem occurs when the passed value of type 'const X*' needs to be pushed to the backing 'std::vector<void*>' – i.e. a non-const 'void*' element type. The same problem exists for push_front(). A simple fix for 'push_back()' (and similar for 'push_front()') is to modify the file 'ptr_sequence_adapter.hpp' adding a cast of the value before pushing is: --- a/include/boost/ptr_container/ptr_sequence_adapter.hpp +++ b/include/boost/ptr_container/ptr_sequence_adapter.hpp @@ -246,7 +246,7 @@ namespace ptr_container_detail this->enforce_null_policy( x, "Null pointer in 'push_back()'" ); auto_type ptr( x ); // notrow - this->base().push_back( x ); // strong, commit + this->base().push_back( (void*) x ); // strong, commit ptr.release(); // nothrow } I am using a C-style cast as neither 'const_cast' nor 'reinterpret_cast' can be used here. (I know the formatting of the comments should also be fixed, but that would make my diff here more difficult to read.) Is there another workaround for this or have I overlooked something? /Martin

2 3

[multi-index] multi-index type trait
by Elizabeta 22 Aug '15

22 Aug '15

Hi how to test if the first index of boost::multi-index is key based or sequence based at compile time. I need to write something like this : template<class T,class M> void f(T& object,typename boost::enable_if<first_index_is_sequence<T>>::type* dummy = 0) { M element; object.push_back(element); } template<class T,class M> void f(T& object,typename boost::enable_if<first_index_is_key<T>>::type* dummy = 0) { M element; object.insert(element); } -- View this message in context: http://boost.2283326.n4.nabble.com/multi-index-multi-index-type-trait-tp467… Sent from the Boost - Users mailing list archive at Nabble.com.

2 5