- Boost-users - lists.boost.org

boost::bind memory leak
by bit bit 11 Nov '10

11 Nov '10

I am using a shared_ptr to execute code on method exit as instructed in http://www.boost.org/doc/libs/1_44_0/libs/smart_ptr/sp_techniques.html#on_b… shared_ptr<void> shutdown(static_cast<void*>(0), bind(&Thread::AnnounceShutdown, this, ref(state_lock))); The above line of code causes a memory leak when an exception is thrown inside AnnounceShutdown(). I know I am not supposed to throw inside a destructor, but in this particular case I see no reason for a memory leak. AnnounceShutDown() does not allocate memory. I suspect ref() has something to do with that. debugger report: WARNING: Visual Leak Detector detected memory leaks! ---------- Block 14928 at 0x00000000006E6FF0: 56 bytes ---------- Call Stack: d:\program files\boost\boost_1_44_0\boost\smart_ptr\detail\shared_count.hpp (124): boost::detail::shared_count::shared_count<void * __ptr64,boost::_bi::bind_t<void,boost::_mfi::mf1<void,bIt::Thread,boost::upgrade_lock<boost::shared_mutex> & __ptr64>,boost::_bi::list2<boost::_bi::value<bIt::Thread * __ptr64>,boost::reference_wrapper<boos d:\program files\boost\boost_1_44_0\boost\smart_ptr\shared_ptr.hpp (199): boost::shared_ptr<void>::shared_ptr<void><void,boost::_bi::bind_t<void,boost::_mfi::mf1<void,bIt::Thread,boost::upgrade_lock<boost::shared_mutex> & __ptr64>,boost::_bi::list2<boost::_bi::value<bIt::Thread * __ptr64>,boost::reference_wrapper<boost::upgrade d:\ptyxiaki\owner\owner\thread\thread.cpp (690): bIt::Thread::Main ... .. . I have already resolved this issue using BOOST_SCOPE_EXIT instead of shared_ptr, but I am looking forward for a better solution with shared_ptr.

1 0

Hashtable
by Ricardo Uribe Lobello 11 Nov '10

11 Nov '10

Good afternoon, I have been trying to use the boost::intrusive::hashtable class but i don't understand clearly how it works. Additionally, I would like to change the hashing function used. Where can i get some example of the hashtable usage? Thanks in advance Ricardo

2 1

[1.45.0] Beta 1 available from SourceForge
by Beman Dawes 11 Nov '10

11 Nov '10

Boost release1.45.0 beta 1 is now available from SourceForge. For those who downloaded release candidate 1, note that beta 1 was build from a slightly later release candidate. See http://sourceforge.net/projects/boost/files/ This release contains numerous enhancements and bug fixes for existing libraries. For details of what's in the release, see http://beta.boost.org/users/news/version_1_45_0. Note that the links to files on this web page are for the final release - use the SourceForge link above to get the beta files. Please download the beta, give it a try, and report any problems you encounter. There are still a few serious bugs outstanding, and we will be working on clearing those before the final release. Thanks, -- The Boost Release Team

6 6

[date_time][1.45] added missing includes (was: [1.45.0] Beta 1 available from SourceForge)
by Eric Niebler 11 Nov '10

11 Nov '10

On 11/10/2010 2:46 PM, Ruediger Berlich wrote: > In order to get it to compile I hat to make a single change: > I had to add "#include <boost/serialization/nvp.hpp>" prior to the inclusion > of the date_time libraries in a single file This seems pretty cut-n-dry. I hope nobody minds, but I went ahead and added the missing #include to gregorian/greg_serialize.hpp and posix_time/time_serialize.hpp on both trunk and release. -- Eric Niebler BoostPro Computing http://www.boostpro.com

1 0

[BGL] Question on dijkstra-example.cpp
by Cedric Laczny 11 Nov '10

11 Nov '10

Hello, could you please explain what this code in the dijkstra-example (http://www.boost.org/doc/libs/1_44_0/libs/graph/example/dijkstra-example.cpp) exactly does? predecessor_map(&p[0]).distance_map(&d[0]) I understand the non-named-parameter version, but the combination of predecessor_map with distance_map is not so obvious to me. What does the "." do there? Do I get it right, that the predecessor_map or distance_map, respectively, automatically wrap the vectors p and d into iterator_property_maps? Thank you. Best regards, Cedric

2 1

[iostreams] Problem with gzip_decompressor
by 陳韋任 11 Nov '10

11 Nov '10

Dear all, Some compressed data was stored in a vector<char> buf (don't know if it's a right way to store compressed data). I used the "buf" as a source, and tried to add a gzip_decompressor to restore those data. I have tried the following approach, but it doesn't work. vector<char> buf; /* fill in the buf */ namespace io = boost::iostreams; io::filtering_istream in(boost::make_iterator_range(buf)); in.push(io::gzip_decompressor()); What I got is an exception, terminate called after throwing an instance of 'boost::exception_detail::clone_impl<boost::exception_detail::error_info_injector <std::logic_error>>' what(): chain complete Aborted How can I compose a source which comes from a container, and add a gzip_decompressor to the source? Thanks in advance. Regards, chenwj -- Wei-Jen Chen (陳韋任) Programming Language and System Laboratory (EC131) Department of Computer Science, National Chiao Tung University

3 4

[fusion] Using fusion::at_c on a sequence or an unused type
by teajay 10 Nov '10

10 Nov '10

Hi, I'm working on finalizing an extension to spirit/qi in which I need to extract an attribute from a fusion sequence in a parser context. The problem I'm facing is that this parser attribute can be an unused type or a valid fusion sequence. Just to illustrate my problem: template <typename Elements,typename Attribute> struct parser_dispachter { template <typename Index> void parse(Index &idx, Attribute &attr) { fusion::at_c<Index::value>(Elements).parse(fusion::at_c<Index::value>(attr)); } }; I actually have two cases: 1. Attribute = unused_type -> at_c replacement should return unused 2. Attribute = fusion sequence -> at_c replacement should return the corresponding attribute reference. What would be the best solution to extract the give attribute ? Regards, Thomas Bernard

2 2

[build] installing layout=versioned into same 'prefix'
by Joe Steeve 10 Nov '10

10 Nov '10

Hello, I am doing this on Debian GNU/Linux. When installing with --layout=versioned, boost.Build creates symlinks without 'version tag' to libraries with version-tag in their name. This, IMHO, defeats the purpose of --layout=versioned. These symlinks collide with existing layout=system installation, in the same prefix. Is there any way to avoid this? Regards, Joe -- Joe Steeve HiPro IT Solutions Pvt. Ltd. http://hipro.co.in/ -- Joe Steeve HiPro IT Solutions Pvt. Ltd. http://hipro.co.in/

1 0

[BFL] Efficient neighbourhood search in trees
by Vincent Spruyt 10 Nov '10

10 Nov '10

Hi all, I'm trying to calculate the average edge weight within a certain neighbourhood for each vertex in a tree. Basically, I have a (minimum spanning) tree and I want to remove all edges whose weight is much larger than the weight of edges closeby. The result would be a forrest of several connected components (I'm working on a clustering algorithm). Currently I naively iterate over each edge. For each edge, I find the source and target vertex. Then, for both the source and the target vertices, I calculate the average weight in a certain neighbourhood using a recursive function as following simplified pseudocode indicates: E: List of edges in MST getNeighbourhoodStats( depth, startvertex, edge ){ out_edges = getIncedentEdges(startvertex) foreach out_edges as out if( out!=edge){ totweight += getWeight(out); if(depth>0) getNeighbourhoodStats(depth-1, target(out), out) } end } main(){ foreach E as e //Get the total weight on the left of the edge int totweight_left=0; getNeighbourhoodStats( 5, source( e), E, totweight_left ) //Get the total weight on the right of the edge int totweight_right=0; getNeighbourhoodStats( 5, target( e), E, totweight_right ) end } However, this approach is rather inefficient as a lot of calculations are redundant (takes about 10ms for 300 edges which is way too slow for the purpose I had in mind). Any ideas for a better solution? Tnx a lot, Kind regards, Vincent

3 4

[Program Options] Positional args in help message
by Lindley M French 10 Nov '10

10 Nov '10

The examples for how to use positional arguments are great and seem to work fine. However, I'm wondering if there's an automatic way to include that information in the help message. In particular, since the tutorial goes to the trouble of telling you how to exclude the positional options from the options help, it would be nice to know what to replace it with. I could manually write a line detailing the usage before outputting the options description, of course, but I was wondering if there was anything in PO to handle this automatically.

3 2