- Boost-users - lists.boost.org

[BGL] Having a cost function as weight map
by Maxime van Noppen 06 Sep '13

06 Sep '13

Hi, I can't find how to do this pretty basic thing : having a function instead of a field to compute the cost of edges for various path finding algorithms. I couldn't find the answer in the documentation and Google was of no help. Did I miss something obvious ? Example : ------------------- #include <boost/graph/dijkstra_shortest_paths.hpp> #include <boost/graph/graph_traits.hpp> #include <boost/graph/adjacency_list.hpp> struct my_edge { double cost() const { return cost_ * 2; } double cost_; }; struct my_node { }; int main() { typedef boost::adjacency_list< boost::vecS, boost::vecS, boost::directedS, my_node, my_edge > graph_t; typedef boost::graph_traits<graph_t>::vertex_descriptor vertex_t; graph_t graph; vertex_t v /*= ... */; using boost::weight_map; // Works : boost::dijkstra_shortest_paths(graph, v, weight_map(boost::get(&my_edge::cost_, graph))); // Doesn't compile : //boost::dijkstra_shortest_paths(graph, v, // weight_map(boost::get(&my_edge::cost, graph))); } ------------------- Thanks ! -- Maxime

4 8

How to build Boost binary libraries on Windows X64 using GCC
by guxiaobo1982 05 Sep '13

05 Sep '13

Hi, We are trying to develop a custom R package, which will utilize Boost and RCPP. We are using R 3.0.1 64bit, and the corresponding RTools3 toolchain, which can be downloaded from CRAN, such as http://cran.at.r-project.org/bin/windows/Rtools/Rtools31.exe. Upon our environment we can build test R packages which with C++ srouces and without any other third-part denpendencies, and can use Eclipse C++ IDE to develop demo C++ programes. When comes to Boost, there is a problem to build the binary libraries such as regex, we found a guide at http://fischerlaender.de/development/using-boost-c-libraries-with-gcc-g-und…, it only tells how to use a pre-built Boost binary, and we have tried the steps at http://stackoverflow.com/questions/5299468/unable-to-build-boost-libraries-… 1. cd to toos\build\v2\engine (there is not a src directory) 2. build.bat gcc 3. put the new bjam.exe and b2.exe into the root the boost 4. execute b2 under root of boost It seems boost still looks for cl (which is from the visual studio toolchain) So do you have a pre-built boost library for windows and gcc, or what's the wright way? Thanks. Xiaobo Gu

1 0

[spirit] Qi rule for GPS NMEA
by Michael Powell 05 Sep '13

05 Sep '13

Hello, I'm digging into a qi rule for GPS NMEA and want to approach it as a kind of 2-pass parser. First pass to extract the message and hex-formatted checksum. Next pass to parse out the actual messages from the message. I believe I comprehend how to go about the second pass. For the first pass, I want to parse $XYZ*HH\r\n. That's the literal '$', and everything between there and the literal '*', followed by the two hex characters, as a complete field, followed by the terminal CR+LF. How do I express the "all characters excluding '*'" for the NMEA message? Then the hex field? Thanks... Regards, Michael Powell

3 9

[BGL] Cannot use vertex indices managed through bundled properties with r_c_shortest_path if EdgeList is listS
by potato_research 05 Sep '13

05 Sep '13

As per the thread topic. Given the following minimal (not-)working example: http://pastebin.com/6zpwCb4X I receive the following compilation error: /usr/local/include/boost/graph/r_c_shortest_paths.hpp:244:50: error: invalid conversion from 'boost::graph_traits<boost::adjacency_list<boost::listS, boost::listS, boost::bidirectionalS, VertexProperty, EdgeProperty> >::vertex_descriptor {aka void*}' to 'int' [-fpermissive] int i_cur_resident_vertex_num = cur_label->resident_vertex; If I change my graph from: adjacency_list<listS, listS, bidirectionalS, VertexProperty, EdgeProperty> to adjacency_list<listS, vecS, bidirectionalS, VertexProperty, EdgeProperty> everything seems to be working fine. However, I need to use listS in my code for iterator validity purposes. Is it possible to use r_c_shortest_paths with a graph with listS EdgeList? If so, how? Thanks. -- View this message in context: http://boost.2283326.n4.nabble.com/BGL-Cannot-use-vertex-indices-managed-th… Sent from the Boost - Users mailing list archive at Nabble.com.

2 7

[chrono] this_thread::sleep_for milliseconds
by Michael Powell 05 Sep '13

05 Sep '13

Hello, I am working on a daemon that involves a sleep_for in the main loop, while worker-threads do the "heavier" lifting in the daemon. I have a couple of signals wired up to respond to terminal, kill or other requests, as per daemonization-guidelines. Question for the list, I gather that thread-sleeps are interrupted when these type of signals occur? Which is fine, I don't need to re-sleep, per se, as some examples seem to suggest. But I do guard a flag in a mutex-critical-section and flag whether to stop running the main loop, which then joins the worker threads. Is this about right as far as daemon, thread/sleep, worker thread approaches are concerned? It seems like it to me, but I thought I should run it past the list to see if there are any pearls of wisdom on the subject. Thank you... Regards, Michael Powell

2 1

[units] coulombs to milliamp-hours
by Michael Powell 05 Sep '13

05 Sep '13

Hello, I am looking for a Coulombs to mAh (milliamp-hours) converter. It's conversion is a ratio I think: 1000 / 3600 (A -> mA, seconds -> hours). Haven't checked yet, but does SI include Coulombs? Or for that matter Amperes, where Coulomb = A s, I think. Thank ye. Regards, Michael Powell

2 5

[multi_index] sorted iteration of values
by Mathieu Champlon 05 Sep '13

05 Sep '13

Hello, I'm trying to use Boost.MultiIndex to store my data. The idea is to have a kind of bidirectional multimap (I initially started with Boost.BiMap) which could model : std::multimap< SomeType*, std::set< int > > left; std::multimap< int, SomeType*> right; I'll need to : . insert obviously . remove all ints based on a SomeType* . remove one given (SomeType*,int) . for a given int iterate on all SomeType* . for a given SomeType* iterate in order on all ints What I have so far is : typedef SomeType* First; typedef int Second; struct Value { First first; Second second; }; boost::multi_index::multi_index_container< Value, boost::multi_index::indexed_by< boost::multi_index::ordered_unique< boost::multi_index::composite_key< Value, boost::multi_index::member< Value, First, &Value::first >, boost::multi_index::member< Value, Second, &Value::second > > >, boost::multi_index::ordered_non_unique< boost::multi_index::member< Value, Second, &Value::second > >, boost::multi_index::ordered_non_unique< boost::multi_index::member< Value, First, &Value::first > > > > data; This works, except for the ints iteration based on a given SomeType* because I need them to be sorted, and they're not, e.g. auto range = data.get< 2 >().equal_range( someTypePtr ); for( auto it = range.first; it != range.second; ++it ) ; does not iterate in a sorted order. Is there a way to achieve this without manually sorting the iterated values ? Thank you ! MAT.

2 5

boost::multiprecision::cpp_int encoding and decoding little endian byte arrays
by Steffen Heil (Mailinglisten) 05 Sep '13

05 Sep '13

Hello I need to decode/encode cpp_int values from/into byte arrays. The encoding must be shortest little endian two's complement representation of the value. For example: 00 = 0 01 = 1 ff = -1 ff:7f = 32767 01:80 = -32767 00:80:00 = 32768 00:80 = -32768 01:80:00 = 32769 ff:7f:ff = -32769 ff:ff:ef:9c:d2:40:43:da:4f:40:6b:06:6a:cd:92:c2:93:bf:a4:5c:c3:f1:29:63:1d = 184467440737095516151844674407370955161518446744073709551615 01:00:10:63:2d:bf:bc:25:b0:bf:94:f9:95:32:6d:3d:6c:40:5b:a3:3c:0e:d6:9c:e2 = -184467440737095516151844674407370955161518446744073709551615 I found a way to archive my goal (see below), but it seems way to complex and too inefficient. For encoding I basically converting the cpp_int to a uint8_t to get 8 bit and then shift the cpp_int by 8, creating a new cpp_int... Twice (once to compute it's length, again to encode it). For decoding I already shift the byte read to the "right" position and "|=" it into the existing cpp_int. Isn't there an easier way access the binary representation of an cpp_int? Regards, Steffen --- #include <iostream> #include <boost/multiprecision/cpp_int.hpp> using namespace std; using namespace boost::multiprecision; size_t binLength( cpp_int value ) { if ( value.is_zero() ) return 1; if ( value.sign() < 0 ) value = ~ value; size_t length = 0; uint8_t lastByte; do { lastByte = value.convert_to<uint8_t>(); value >>= 8; length ++; } while ( ! value.is_zero() ); if ( lastByte >= 0x80 ) length ++; return length; } void binEncode( cpp_int value, uint8_t* output, size_t length ) { if ( value.is_zero() ) *output = 0; else if ( value.sign() > 0 ) while ( length -- > 0 ) { *( output ++ ) = value.convert_to<uint8_t>(); value >>= 8; } else { value = ~ value; while ( length -- > 0 ) { *( output ++ ) = ~ value.convert_to<uint8_t>(); value >>= 8; } } } cpp_int binDecode( uint8_t* input, size_t length ) { boost::multiprecision::cpp_int result( 0 ); int bits = - 8; while ( length -- > 1 ) result |= (boost::multiprecision::cpp_int) *( input ++ ) << ( bits += 8 ); uint8_t a = *( input ++ ); result |= (boost::multiprecision::cpp_int) a << ( bits += 8 ); if ( a >= 0x80 ) result |= (boost::multiprecision::cpp_int) - 1 << ( bits + 8 ); return result; } void test( string s ) { cpp_int x( s ); uint8_t buffer[ 100 ]; size_t length = binLength( x ), index = 0; binEncode( x, buffer, length ); cpp_int y = binDecode( buffer, length ); cout << hex << setw( 2 ) << setfill( '0' ) << (int) buffer[ 0 ]; while ( -- length > 0 ) cout << ':' << setw( 2 ) << setfill( '0' ) << (int) buffer[ ++ index ]; cout << " = " << dec << setw( 1 ) << x << endl; if ( x != y ) cerr << "failed" << endl; } int main( int argc, const char* argv[] ) { test( "0" ); test( "1" ); test( "-1" ); test( "32767" ); test( "-32767" ); test( "32768" ); test( "-32768" ); test( "32769" ); test( "-32769" ); test( "184467440737095516151844674407370955161518446744073709551615" ); test( "-184467440737095516151844674407370955161518446744073709551615" ); }

2 1

Re: [Boost-users] Why fusion::map<> types not the same?
by Sergey Zhuravlev 03 Sep '13

03 Sep '13

This is bug. You should add ticket to bugtracker. Try to use boost\fusion\result_of::as_map as workaround.

2 1

[Log] How to use a channel that is constant per file?
by Timo Schmiade 03 Sep '13

03 Sep '13

Hi all, I want to use boost's log library to output messages to the console which feature a per-message severity and a per-file channel. I already tried two approaches, none of which was satisfying: === 1. approach === The class boost::log::sources::severity_channel_logger supports both severity and channel and I could define it in each source file like so: --- 8< --- BOOST_LOG_INLINE_GLOBAL_LOGGER_DEFAULT( log, boost::log::sources::severity_channel_logger< boost::log::trivial::severity_level, std::string > ) --- >8 --- Unfortunately, using it like this requires me to redundantly add the channel to each log message, which is very inconvenient because it is the same for all logs in the file. --- 8< --- // Message.cpp BOOST_LOG_CHANNEL_SEV(log, Message, trace) << "Hello, world"; --- >8 --- === 2. approach === I defined my logger like this: --- 8< --- BOOST_LOG_INLINE_GLOBAL_LOGGER_DEFAULT( log, boost::log::sources::severity_logger< boost::log::trivial::severity_level > ) --- >8 --- Then, in each file, I manually set the "Channel" attribute for the logger, like so: --- 8< --- // Message.cpp log::get().add_attribute( "Channel", boost::log::attributes::make_constant("Message")); --- >8 --- This works, but as the "log" object seems to be stored globally, I would have to "invent" new names for the logger object in each file. This is both insecure and prevents the usage of macros (I think). So here's the question: What would be the recommended / a convenient way to add a per-file constant channel to log messages? Thanks in advance! Timo

2 2