October 2009 - Boost-users - lists.boost.org

newbie: help to start with object_pool
by Jenia 08 Oct '09

08 Oct '09

Hi this is my first try with boost. I need an object pool, where I can preallocate X objects and reuse them with ability to allocate more if the pool runs empty and shrink my pool if needed. My questions: 1. Is object_pool suitable for this purpose? 2. Does object_pool.free() returns object memory to the pool or return pool's memory to the system? 3. How do I use object_pool.construct()? I didn't get the number of parameters requirement in the documentation (http://www.boost.org/doc/libs/1_40_0/libs/pool/doc/interfaces/object_pool.h…)

2 1

Re: [Boost-users] error C2733 second C linkage of overloaded function InterlockedIncrement not allowed
by Darth Vader 08 Oct '09

08 Oct '09

This is an old thread (1 year old) I'm answering to: This is about compiling some of the Boost library (1.39) on Windows CE 6.0. I got the following error: error C2733 second C linkage of overloaded function _InterlockedIncrement not allowed It turns out that "_InterlockedIncrement" is already defined in winbase.h so I commented out: extern "C" long __cdecl InterlockedIncrement( long* ); extern "C" long __cdecl InterlockedDecrement( long* ); extern "C" long __cdecl InterlockedCompareExchange( long*, long, long ); extern "C" long __cdecl InterlockedExchange( long*, long ); extern "C" long __cdecl InterlockedExchangeAdd( long*, long ); In interlocked.hpp and everything compiled fine.

1 0

[xpressive] Calling a placeholder method from within an action?
by Ami Ganguli 08 Oct '09

08 Oct '09

Hi all, I'm using an object as a placeholder, and would like to call various methods to the object from within my actions. Something along the lines of: port = (s1= *digit) [ _placeholder->setPort( as<int>(s1) ) ]; I've found a way to do something equivalent using method pointers, but the resulting code is really opaque. I'm afraid nobody else will ever be able to figure out what I've done. I also know I can do this by creating separate function objects for each method that will be called. This is less opaque, but really verbose. I end up with hundreds of lines of boilerplate code. Is there a "right" way to achieve something similar? Regards, Ami.

2 2

Stripped down "runtime" version of boost
by Paul Moore 08 Oct '09

08 Oct '09

At the moment, I'm using boost by simply unpacking the archive (I've only needed the header-only libraries thus far). But it's a fairly large archive, full of things that aren't actually needed to simply *use* the libraries. Is there a means of creating a subset of the boost distribution with *just* the headers (and libraries, ultimately) needed to compile programs using boost? I'm guessing that all I'd need is the boost subdirectory (with the headers) plus whatever parts of the lib directory are created as part of the boost.build process. Is that right? Are there any more explicit instructions anywhere? Thanks, Paul.

3 2

BGL -- adjacency_matrix, graph concepts
by Raymond Wan 08 Oct '09

08 Oct '09

Hi all, I have a basic question about BGL, but I'm a bit too new with templating to understand some of the terms used in the documentation... :-( I'd like to build a graph using an adjacency matrix and the edge_iterator_constructor of the graph (or anything faster than repeated add_edge's). In the docs, I found this for adjacency_list: template <class EdgeIterator> adjacency_list(EdgeIterator first, EdgeIterator last, vertices_size_type n, edges_size_type m = 0, const GraphProperty& p = GraphProperty()) but the closest I found for adjacency_matrix is: template <typename EdgeIterator> adjacency_matrix(EdgeIterator first, EdgeIterator last, vertices_size_type n, const GraphProperty& p = GraphProperty()) It's not the same because of "typename"? As a toy example, I've taken the MST example and have modified it a lot to get this: ----- #include <boost/config.hpp> #include <boost/graph/adjacency_matrix.hpp> #include <boost/graph/graph_utility.hpp> #include <iostream> int main () { using namespace boost; typedef adjacency_matrix < undirectedS, no_property, property < edge_weight_t, int > > Graph; typedef graph_traits < Graph >::edge_descriptor Edge; typedef graph_traits < Graph >::vertex_descriptor Vertex; typedef std::pair<int, int> E; const int num_nodes = 5; E edges[] = { E(0, 2), E(1, 3), E(1, 4), E(2, 1), E(2, 3), E(3, 4), E(4, 0) }; int weights[] = { 1, 1, 2, 7, 3, 1, 1 }; std::size_t num_edges = sizeof(edges) / sizeof(E); Graph g1 (num_nodes); for (unsigned int i = 0; i < num_edges; i++) { add_edge (edges[i].first, edges[i].second, g1); } print_graph (g1); Graph g2 (num_nodes); // Graph g2 (edges, edges + num_edges, weights, num_nodes); print_graph (g2); return EXIT_SUCCESS; } ----- The 5th line from the end is what is causing me problems. It seemed to work with an adjacency_list. (I'd also would like to add weights, but I guess add_edge could support that...I just think that this way is faster based on the "Quick Tour of the BGL" page in the docs.) I looked in my examples directory [1.39] and simply grepped for "adjacency_matrix" and came up with only one file: adjacency_matrix.cpp . Other than the pros/cons of adjacency matrices and lists, is there something wrong with adjacency matrices; any reason for its absence in the examples directory? One final question -- in the documentation, it lists various graph concepts (http://www.boost.org/doc/libs/1_40_0/libs/graph/doc/graph_concepts.html) such as BidirectionalGraph, VertexAndEdgeListGraph, AdjacencyMatrix, etc. Are these concepts relevant to BGL or only in a general sense? What I mean is what is their relationship with adjacency_list and adjacency_matrix? Any pointers would be appreciated... Thank you! Ray

3 11

[lambda][shared_ptr] Creating a shared_ptr with a custom delete function using a lambda expression
by Ryan McConnehey 07 Oct '09

07 Oct '09

I have a class that maintains a list of available numbers. The user will ask for a number and receive a shared_ptr with a value. Once the user is done with the number it will be put back on the list of available numbers. I'm trying to initialize the list with numbers and a custom delete function for the shared_ptr to be return to the list. Any help would be appropriated in how to get the lambda and custom delete function to work. Ryan class CTNGenerator { public: CTNGenerator(void) { m_AvailableCTNs.resize(20); unsigned short startValue = 15; typedef boost::shared_ptr<unsigned short> bus; //The lambda function isn't compiling with how it's currently written. std::for_each(m_AvailableCTNs.begin(), m_AvailableCTNs.end(), _1 = bus(new bus(var(startValue++)), &makeCTNAvailable)); } private: void CTNGenerator::makeCTNAvailable(unsigned short * ctn) { m_AvailableCTNs.push_back(boost::shared_ptr<unsigned short>(ctn)); } private: std::list< boost::shared_ptr<unsigned short> > m_AvailableCTNs; };

1 0

error in recreation of shared memory using boost:: interprocess
by manish4gupta 07 Oct '09

07 Oct '09

Hi, I am creating shared memory with given name(by given file name with key value ) and creating the map in the shared region by parsing key-value from the file. First time it is ok. but when i try to do it again, I am getting following error. terminate called after throwing an instance of 'boost::interprocess::interprocess_exception' what(): File exists Aborted What is the cause of this error. Thanks in advance. -- View this message in context: http://www.nabble.com/error-in-recreation-of-shared-memory-using-boost%3A%3… Sent from the Boost - Users mailing list archive at Nabble.com.

2 1

Setting QOS doesn't work?
by Dominique Lacerte 07 Oct '09

07 Oct '09

Hi, I have some QOS code in a product that is working fine. We're porting that code to using the socket support in ASIO and the QOS code doesn't seem to be working any longer. We're using WinXP under both environments and running our old code works fine but the new code has issues. Setting the TOS bits via: ::setsockopt(mSocket.native(), IPPROTO_IP, IP_TOS, (char *)(&uiTOS), sizeof(uiTOS)) Works fine, but setting the service type does not. We're making the following call: DWORD dwBytesReturned = 0; bool bSuccess = (SOCKET_ERROR != ::WSAIoctl( mSocket.native(), SIO_SET_QOS, (void *)reinterpret_cast<const BYTE *>(&tQOS), sizeof(tQOS), NULL, 0, &dwBytesReturned, NULL, NULL )); With values such as: SocketAddress bytes: 0200fc22c0a863670000000000000000 ===RX=== TokenRate: -1 TokenBucketSize: -1 PeakBandwidth: -1 Latency: -1 DelayVariation: -1 ServiceType: 6 MaxSduSize: -1 MinimumPolicedSize: -1 ===TX=== TokenRate: 4600 TokenBucketSize: -1 PeakBandwidth: -1 Latency: -1 DelayVariation: -1 ServiceType: 2 MaxSduSize: -1 MinimumPolicedSize: -1 And it returns success, but the DSCP bits in the packets are not being set. If we call SIO_GET_QOS afterwards we always get back the default: SocketAddress bytes: <NULL> ===RX=== TokenRate: -1 TokenBucketSize: -1 PeakBandwidth: -1 Latency: -1 DelayVariation: -1 ServiceType: 1 MaxSduSize: -1 MinimumPolicedSize: -1 ===TX=== TokenRate: -1 TokenBucketSize: -1 PeakBandwidth: -1 Latency: -1 DelayVariation: -1 ServiceType: 1 MaxSduSize: -1 MinimumPolicedSize: -1 Is there something different about the sockets implementation used by ASIO that prevents the WSAIoctl() call from working correctly? Dominique Lacerte Senior Software Engineer CounterPath Corporation T 250.380.0584 x1368 M 250.595.1362 [cid:image001.png@01CA472D.8E379C30]<http://www.counterpath.com/>

1 0

[BGL] premature termination in dijkstra using visitor
by Ralf Goertz 07 Oct '09

07 Oct '09

Hi, I have some very large undirected graphs (half a million vertices). Edges have weight 0 or 1. For each vertex I need to find all vertices * that are in sum 2 edges away * whose distance is 0 Both tasks can be done with breadth-first searches but I don't know in advance how many edges in depth I have to traverse. That's why I thought of Dijkstra's algorithm. But in my case this is overkill as I am not interested in the exact distance of a vertex from a given vertex but only in if it is not farther than 0 or 2 away. Is it possible to use a visitor to alter the colormap to mark vertices as final that are too far even thought they are not final yet? Ralf

5 12

[proto] Question on grammar and operator overloading
by Hal Finkel 06 Oct '09

06 Oct '09

Hello, I'm writing a complex number class using boost::proto by extending proto::terminal< boost::array<_, 2> >. I've attached a copy of my current code, which compiles, runs, and (at least with both g++ and icc on my machine) demonstrates a significant speed improvement compared to std::complex. However, there is (at least) one thing which don't work as it should: In the grammar I have to specify lazy_complex< double > instead of lazy_complex< _ >, otherwise the compiler cannot find proto's operator overloads, why? Note: This code will not compile under 1.37, but will compile under 1.40, because of some issue in 1.37 which manifests itself as a problem with the overloading of the '<<' operator. I've not tested the releases in between 1.37 and 1.40. General suggestions are also welcome. I'd like to get this cleaned up for general redistribution. Thank you in advance, Hal

3 7