- Boost-users - lists.boost.org

Re: [Boost-users] boost::function performance issues with C++11
by Seeger, Steven D. (GSFC-444.0)[Embedded Flight Systems, Inc] 08 Jul '14

08 Jul '14

>Right, there are 1000 integers in the vector. If something, say STL, is bottleneck for boost::bind with C++ 11, it >should also be a bottleneck without C++11. And they should have the same performance. However, the fact shows >that C++11 mode has some additional overhead. I wanna figure that out. What if you try with just 1 integer? Does the difference in time stay proportionally the same? Steven

4 4

[spirit] qi xml parser
by Michael Powell 07 Jul '14

07 Jul '14

Hello, I am building out a general use xml parser including attributes, arbitrary number of elements, and so on. So far so good, makes sense parsing names and so forth. However, how do you handle element content? Which could either be a string, or zero or more other elements (basically of the same rule as the enclosing element rule). It would seem you need a terminus, the empty element tag. In such a way that populates the parent (initial) element, and its children (of the same element kind). I'll be adapting structs to capture the results. I am also using a couple of helpful references, for instance: http://www.w3.org/TR/xml11/ http://stackoverflow.com/questions/9473843/boost-spirit-how-to-extend-xml-p… Also not sure quite how to capture the adapted parts at strategic rule opportunities. My domain model will look something like this, keeping it simple as possible: struct xattribute { std::string name; std::string value; }; typedef std::vector<xattribute> xattribute_vector; struct xelement; typedef std::vector<xelement> xelement_vector; struct xelement { std::string name; std::string content; xattribute_vector attributes; xelement_vector children; }; Thanks... Best regards, Michael Powell

2 6

DateTime without serialization
by Macumber, Daniel 07 Jul '14

07 Jul '14

Hello, I am using the DateTime library's to/from string functionality but not the DateTime serialization functionality. However, when I build my program boost auto linking keeps trying to link against the serialization library. Is there a way I can build and link against DateTime without auto linking against serialization? I am using visual studio 2013 and boost 1.55. Thanks, Dan Sent with Good (www.good.com)

3 2

BOOST_OS_UNIX not working
by Troy Heron 07 Jul '14

07 Jul '14

Hello, I've just decided to give Boost.Predef a test drive for the first time and have found a problem using BOOST_OS_UNIX. See the test snippet here: http://pastebin.com/4dLDAhEV This is the output I get using both g++ 4.9.0 and clang++ 3.4.2: BOOST_OS_UNIX: 0 BOOST_OS_LINUX: 1 unix: 1 __unix: 1 _XOPEN_SOURCE: 700 _POSIX_SOURCE: 1 Everything that is defined that should be to make BOOST_OS_UNIX be defined. I haven't seen this mentioned anywhere before. I guess this a bug? I haven't looked at predef headers yet. Thanks, Troy

2 1

How to Wrap a List of pairs with Python Boost
by Kv Gopalkrishnan 07 Jul '14

07 Jul '14

Hi all I am quiet new to python boost and am stuck with a problem with C++ STL lists. I have a data type that is something like this list<pair<int,string> > gap_ls; Which is under the private section. I would like to wrap this like one can do using a vector like this class_<std::vector<std::pair<double, double> > >("VectorOfPairs") .def(vector_indexing_suite<std::vector<std::pair<double,double> > >()); and for pairs I had this class_<std::pair<double, double> >("DoublePair") .def_readwrite("first", &std::pair<double, double>::first) .def_readwrite("second", &std::pair<double, double>::second); But how do I write a wrapper for list<pair<int,string> >gap_ls; Any advice in this context would be appreciated. Kind Regards K.V.

1 0

Re: [Boost-users] Boost-users Digest, Vol 3844, Issue 1
by Seeger, Steven D. (GSFC-444.0)[Embedded Flight Systems, Inc] 05 Jul '14

05 Jul '14

>Date: Sun, 29 Jun 2014 13:21:21 -0300 >From: Agust?n K-ballo Berg? <kaballo86(a)hotmail.com> >To: boost-users(a)lists.boost.org >Subject: Re: [Boost-users] [fusion] maps and arrays >Message-ID: <BLU436-SMTP192734464D2B326EAF70973A7050(a)phx.gbl> >Content-Type: text/plain; charset="ISO-8859-1"; format=flowed >I missed the part where `for_each` was invoking the callable on each >element of the array, that is not how it is supposed to be. You are >right that making a recursive callable is the correct way to obtain that >behavior. Hi Agustin I'm sorry for taking so long to reply to you. I've been swamped. I really appreciate you taking the time to have this discussion with me and helping me to see that the behavior I was experiencing before with the for_each stepping into each array item is not correct. I should remember in the future to try and understand from the documentation what something is doing and not just by empirically testing it. >Without any other information on the error you get, I'm betting that >this is due to adding support for C++11 (not-so) perfect forwarding: > void foo(std::array<int, 2> const&){} > template <typename T> > void bar(T&& v){ foo(std::forward<T>(v)); } > foo({3,4}); // ok > bar({3,4}); // error: {3,4} has no type If I get some time I will play with this. I think that it is a bug that I cannot use braced initializer in c++11 with 1.55 like I could in 1.53. In the meantime I am adding a comment and referencing this discussion to my bug report to show that it is not a bug. I imagine they are not interested in the 1.53 issue we may have uncovered. Regards, Steven

1 0

skip list in boost
by David Frank 05 Jul '14

05 Jul '14

Hi, I found this email thread about submitting skip list implementation to boost: Boost mailing page: Re: [boost] Review managers wanted for augmented data structures Boost mailing page: Re: [boost] Review managers wanted f... Boost : Date view Thread view Subject view Author view Subject: Re: [boost] Review managers wanted for augmented data structures From: Vadim Stadnik (vadimst... View on lists.boost.org Preview by Yahoo But, I do not see it in the current version of boost, I'm wondering if it is somewhere I missed? Thanks, David

2 2

[Graph] Newbie on A*
by Sensei 04 Jul '14

04 Jul '14

Dear all, I am fiddling with the A* example on cities in order to learn how to use the A* search. Obviously, I cannot make it work :) The main goal is this, but I'm way too far from this: make A* fail when the path is longer, **not costlier**, than a constant. But right now I'd just like to have A* working on a graph. :) In order to see the different results, I've paired it with a Dijkstra search, and my A* heuristics is quite simple: constant. Dijkstra works fine, but A* stops at the first node, and the target isn't reached (but a path is available). What am I doing wrong? Forgive me if I'm making a stupid mistake bu I'm completely a newbie! If you need the graph, here's the link: https://dl.dropboxusercontent.com/u/15635416/g.txt Thanks! #include <cstdio> #include <cstdlib> #include <iostream> #include <utility> #include <algorithm> #include <numeric> #include <deque> #include <unordered_set> #include <vector> #include <fstream> #include <list> #include <boost/graph/graph_traits.hpp> #include <boost/graph/compressed_sparse_row_graph.hpp> #include <boost/graph/dijkstra_shortest_paths.hpp> #include <boost/graph/breadth_first_search.hpp> #include <boost/graph/depth_first_search.hpp> #include <boost/pending/indirect_cmp.hpp> #include <boost/range/irange.hpp> #include <boost/property_map/property_map.hpp> #include <boost/graph/astar_search.hpp> // Properties: weights are int typedef boost::property<boost::edge_weight_t, int> edge_weight; // The graph itself as a compressed sparse row matrix typedef boost::compressed_sparse_row_graph<boost::bidirectionalS, boost::no_property, edge_weight> boost_graph; // Vertex iterator typedef boost::graph_traits<boost_graph>::vertex_iterator vertex_iterator; // Edge iterator typedef boost::graph_traits<boost_graph>::edge_iterator edge_iterator; // Adjacent nodes iterator typedef boost::graph_traits<boost_graph>::adjacency_iterator adjacency_iterator; typedef boost::graph_traits<boost_graph>::out_edge_iterator outward_iterator; typedef boost::graph_traits<boost_graph>::in_edge_iterator inward_iterator; // Property maps typedef boost::property_map<boost_graph, boost::vertex_index_t>::type index_map; //typedef boost::property_map<boost_graph, boost::vertex_name_t> name_map; // Property iterators typedef boost::iterator_property_map<std::size_t*, index_map, std::size_t, std::size_t&> predecessor_map; typedef boost::iterator_property_map<std::size_t*, index_map, std::size_t, std::size_t&> distance_map; bool isValid(std::size_t index) { return (index >= 0) && (index <= 12); } // **************************************************************************** // A* heuristics template <class graph_type> class astar_heuristic : public boost::astar_heuristic<graph_type, std::size_t> { public: typedef typename boost::graph_traits<graph_type>::vertex_descriptor vertex_type; astar_heuristic(vertex_type goal) : goal_(goal) { }; std::size_t operator()(vertex_type u) { return 3; } private: vertex_type goal_; }; // Class to be used to terminate the A* algorithm class exit_astar { public: exit_astar(bool found) { }; }; // A* visitor that terminates when we find the goal template <class vertex_type> class astar_visitor : public boost::default_astar_visitor { public: astar_visitor(vertex_type goal) : goal_(goal) { }; template <class Graph> void examine_vertex(vertex_type u, Graph& g) { //std::cout << "EXAM: " << u << std::endl; if (u == goal_) throw exit_astar(true); } private: vertex_type goal_; }; // **************************************************************************** int main(int argc, const char * argv[]) { std::vector<std::pair<std::size_t, std::size_t>> graph_edges; std::vector<int> edge_weight; int n_nodes; int read_a, read_b, read_w; n_nodes = -1; std::ifstream igraph("/Users/senseiwa/Documents/Projects/scratch/tbb/g.txt"); while(igraph >> read_a >> read_b >> read_w) { if (read_a > n_nodes) n_nodes = read_a; if (read_b > n_nodes) n_nodes = read_b; read_a--; read_b--; graph_edges.push_back(std::make_pair(read_a, read_b)); edge_weight.push_back(read_w); } std::cout << "+++ READ # NODES: " << n_nodes << std::endl; // Create the graph boost_graph graph(boost::edges_are_unsorted_multi_pass, graph_edges.begin(), graph_edges.end(), edge_weight.data(), n_nodes); std::cout << "Graph statistics: " << std::endl; std::cout << "Nodes: " << boost::num_vertices(graph) << std::endl; std::cout << "Edges: " << graph_edges.size() << std::endl; edge_iterator ei, ee; std::size_t source_node = std::stoul(argv[1]); std::size_t destination_node = std::stoul(argv[2]); std::cout << ">> SOURCE NODE " << source_node << std::endl; std::cout << "Degrees in/out: " << source_node << std::endl; std::cout << boost::in_degree(boost::vertex(source_node, graph), graph) << std::endl; std::cout << boost::out_degree(boost::vertex(source_node, graph), graph) << std::endl; std::cout << "Degrees in/out: " << destination_node << std::endl; std::cout << boost::in_degree(boost::vertex(destination_node, graph), graph) << std::endl; std::cout << boost::out_degree(boost::vertex(destination_node, graph), graph) << std::endl; std::vector<std::size_t> preds(boost::num_vertices(graph)); std::vector<std::size_t> dists(boost::num_vertices(graph)); index_map idx_map = boost::get(boost::vertex_index, graph); predecessor_map pred_map(preds.data(), idx_map); distance_map dist_map(dists.data(), idx_map); boost::dijkstra_shortest_paths(graph, source_node, boost::distance_map(dist_map).predecessor_map(pred_map)); typedef std::vector<boost_graph::edge_descriptor> path_type; path_type spath; std::cout << ">> SOURCE NODE " << source_node << std::endl; std::cout << ">> DESTINATION NODE " << destination_node << std::endl; // ******* DIJKSTRA ******************************************************** std::size_t v = destination_node; // We want to start at the destination and work our way back to the source for(std::size_t u = pred_map[v]; // Start by setting 'u' to the destintaion node's predecessor u != v; // Keep tracking the path until we get to the source v = u, u = pred_map[v]) // Set the current vertex to the current predecessor, and the predecessor to one level up { std::pair<boost_graph::edge_descriptor, bool> edgePair = boost::edge(u, v, graph); boost_graph::edge_descriptor edge = edgePair.first; spath.push_back(edge); } // Write shortest path int totalDistance = 0; for(path_type::reverse_iterator pathIterator = spath.rbegin(); pathIterator != spath.rend(); ++pathIterator) { std::cout << boost::source(*pathIterator, graph) << " - " << boost::target(*pathIterator, graph) << " = " << boost::get(boost::edge_weight, graph, *pathIterator) << std::endl; totalDistance += boost::get(boost::edge_weight, graph, *pathIterator); } std::cout << "Shortest Dijkstra path from " << source_node << " to " << destination_node << " cost " << totalDistance << std::endl; // ******* A* ************************************************************** // A* search predecessor and cost std::vector<std::size_t> p(boost::num_vertices(graph)); std::vector<std::size_t> d(boost::num_vertices(graph)); try { // A* parameter interface astar_search(graph, source_node, astar_heuristic<boost_graph>(destination_node), boost::predecessor_map(&p[0]).distance_map(&d[0]).visitor(astar_visitor<std::size_t>(destination_node))); } catch(exit_astar fg) { std::list<std::size_t> shortest_path; for(std::size_t v = source_node; ; v = p[v]) { shortest_path.push_front(v); if(p[v] == v) break; } auto spi = shortest_path.begin(); std::cout << "node " << source_node << std::endl; for(++spi; spi != shortest_path.end(); ++spi) std::cout << "node " << *spi << std::endl; std::cout << "Shortest A* path from " << source_node << " to " << destination_node << " cost " << d[destination_node] << std::endl; std::cout << shortest_path.size() << std::endl; return 0; } std::cout << "Didn't find a path from " << source_node << "to" << destination_node << "!" << std::endl; return 0; } +++ READ # NODES: 200 Graph statistics: Nodes: 200 Edges: 4000 >> SOURCE NODE 1 Degrees in/out: 1 0 35 Degrees in/out: 192 39 1 >> SOURCE NODE 1 >> DESTINATION NODE 192 1 - 47 = 2 47 - 91 = 3 91 - 120 = 1 120 - 175 = 1 175 - 192 = 10 Shortest Dijkstra path from 1 to 192 cost 17 node 1 Shortest A* path from 1 to 192 cost 17 1

2 2

Re: [Boost-users] Boost-users Digest, Vol 3847, Issue 2
by Jorge Bellon 04 Jul '14

04 Jul '14

Buenos días Vicenç, Te envío la respuesta del desarrollador de la librería. Tiene sentido lo que dice. La única pega, quizás, sería que, en caso de no haber memoria disponible, tardaríamos al menos tanto tiempo como el "round-trip latency" antes de darnos cuenta... Un saludo, Jorge > Message: 1 > Date: Thu, 03 Jul 2014 20:08:33 +0200 > From: Ion Gazta?aga <igaztanaga(a)gmail.com> > To: boost-users(a)lists.boost.org > Subject: Re: [Boost-users] [Interprocess] Managed segment placement > Message-ID: <53B59C21.9040901(a)gmail.com> > Content-Type: text/plain; charset=UTF-8; format=flowed > > El 03/07/2014 10:24, Jorge Bellon escribi?: > > Hi Ion, > > Thank you for the info. > > > > The use case would be to manage a memory pool allocated on a remote node. > > > > This would be used by a run-time library for distributed programming. It > > would allocate a memory region and manage this memory from another node, > > so that we can perform operations with that piece of memory (allocate > > portions, transfer data, etc.) without the intervention of the local > > node. > > Why can't you put the manager in the remote node? The manager also needs > to allocate memory for indexes that track the position of named/unique > objects and that memory comes from the memory region also used to > construct objects. If you don't need named objects you can use a null > index to minimize used memory. > > Do you need to shared this remote node memory between processes? If you > only use remote memory from a single process you can speed things a bit > using raw pointers: > > typedef basic_managed_external_buffer < > char, > rbtree_best_fit<null_mutex_family, void* >, > null_index > > > my_managed_external_buffer; > > If you want to serialize objects then you should use offset_ptr<void> > instead of void*. > > I still miss the point of separating the manager (which is placed in the > beginning of the memory region) and the rest of the region. > > Best, > > Ion WARNING / LEGAL TEXT: This message is intended only for the use of the individual or entity to which it is addressed and may contain information which is privileged, confidential, proprietary, or exempt from disclosure under applicable law. If you are not the intended recipient or the person responsible for delivering the message to the intended recipient, you are strictly prohibited from disclosing, distributing, copying, or in any way using this message. If you have received this communication in error, please notify the sender and destroy and delete any copies you may have received. http://www.bsc.es/disclaimer

1 0

Re: [Boost-users] [atomic] ThreadSanitizer reports a data race when using the code from "reference counting" example
by Adam Romanek 04 Jul '14

04 Jul '14

On 01.07.2014 10:49, Andrey Semashev wrote: > On Tue, Jul 1, 2014 at 12:41 PM, Adam Romanek wrote: >> On 01.07.2014 10:36, Andrey Semashev wrote: >>> >>> On Tue, Jul 1, 2014 at 12:20 PM, Andrey Semashev >>> wrote: >>>> >>>> On Tue, Jul 1, 2014 at 11:20 AM, Adam Romanek >>>> wrote: >>>>> >>>>> Andrey, >>>>> >>>>> If you don't mind, please have a look at the conversation that I started >>>>> on >>>>> boost-users list [1] a few days back. Since you're the maintainer of >>>>> Boost.Atomic I'd like to know your opinion on this issue. Or maybe I >>>>> should >>>>> repost this on boost-dev list? >>>> >>>> >>>> I don't follow users list, mostly because I can't even keep up with >>>> the developers list. So it's better to reach me on the dev list. >>>> >>>> The code in the StackOverflow topic indeed contains a potential >>>> problem. The fetch_sub(release) operation prevents previous operations >>>> from sinking past it and atomic_thread_fence(acquire) prevents the >>>> following operations from floating above it. However, these two >>>> barriers themselves can be reordered creating a window where >>>> operations before fetch_sub(release) and after >>>> atomic_thread_fence(acquire) may overlap. I believe, this is permitted >>>> on both the compiler and hardware levels. This doesn't mean that's >>>> what is actually happening and being flagged by TSan; it is still >>>> possible that some TSan bug is exposed. The fact that replacing >>>> atomic_thread_fence(acquire) with fetch_add(acquire) speaks in favor >>>> of the bug suspicion. >>> >>> >>> Actually, no, fetch_add(acquire) is significantly different from the >>> fence. It operates on the same object as the previous >>> fetch_sub(release), so it must observe its effects and cannot be >>> reordered. >>> >> >> After reading your first response I thought you're suggesting that the >> "reference counting" example in Boost.Atomic docs is broken. But hopefully >> we agree that it's fine, right? :) > > The example in the docs is indeed not correct. I believe, it should be > corrected to use acq_rel memory order in intrusive_ptr_release. The > fence can be removed then. I'll fix it as soon as I have some time. > >> So I guess this is a false positive from TSan, as suggested by others on SO. >> I'll ask this on TSan's mailing list, lets see what they say. > > Ok, let me know if there are other considerations from TSan devs. > See above for my private conversation with Andrey Semashev on this topic (he does not follow the boost-users mailing list). Andrey says that the "reference counting example" from Boost.Atomic docs is broken. What do you think about this? WBR, Adam Romanek

2 2