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

Recent compilation troubles with gcc atomics
by tom fogal 02 May '09

02 May '09

Hello all, Recently I have noted more than one posting to this list with regards to compilation troubles on some platforms utilizing Gnu C++ or Intel's C++ compiler. Specifically, I am referring to linking errors with some of gcc's internal `atomics' functions, e.g. `__sync_fetch_and_add'. Some searching brings up: http://www.mail-archive.com/debian-bugs-closed@lists.debian.org/msg197119.h… http://www.nabble.com/Building-for-ARM926EJ-S-td16447579.html http://groups.google.com/group/boost-list/browse_thread/thread/8e757cc2788d… https://svn.boost.org/trac/boost/ticket/2525 http://www.nabble.com/issues-with-unit-tests-on-intel-32-bit,-gcc-4.3,-unde… I also happen to follow the libstdc++ mailing list, which has just seen an elucidating post by Joseph S. Myers: http://gcc.gnu.org/ml/gcc/2009-05/msg00039.html In it, he notes that the availability of these builtins is not solely a function of the compiler version. In particular, the platform (CPU, OS, runtime support) seems to be a factor, as well as the backend code generation options (i.e. -march gcc compiler option). He also suggests that the only reliable methodology for discovering the availability of these functions/builtins is by doing link-time tests. I mention as much because (unverified) I think boost's current solution is #ifdefs based on architecture and compiler version, which I think is fighting a losing battle. I hope this information is useful for those stuck on the problem, and/or boost developers. Cheers, -tom

1 0

[asio] async write handler is not called until the socket is not closed
by Roman Shmelev 02 May '09

02 May '09

Hi! I've got very strange problem - handler passed to async_write is not called in time. What I mean - it is called only when socket is closed - then handler is invoked many times. Sometimes it is so, sometimes everything is fine. The main thing is that traffic exists. I mean - async writes get completed. I call io_service.run() in proper time, etc. Another moment - the first two times the handler is invoked, but not more. As I guess, the problem is somewhere in my code. But I ask you to point to probable reasons of such behaviour that breaks asio rules..

2 7

[iterator] std::sort on zip_iterator
by Jeremiah Willcock 02 May '09

02 May '09

I would like to sort two separate vectors of numbers, treating corresponding elements of the two vectors as tuples and sorting the tuples. In particular, I am only comparing the first elements of the tuples; that may or may not be relevant to my problem. An example program is: #include <vector> #include <algorithm> #include <iostream> #include <boost/tuple/tuple.hpp> #include <boost/iterator/zip_iterator.hpp> struct compare_first_elements_in_tuples { template <typename Tuple> bool operator()(const Tuple& a, const Tuple& b) const { return (a.template get<0>()) < (b.template get<0>()); } }; int main(int, char**) { size_t a_data[5] = {0, 2, 1, 2, 4}; size_t b_data[5] = {1, 1, 5, 3, 7}; std::vector<size_t> a(a_data, a_data + 5); std::vector<size_t> b(b_data, b_data + 5); std::cout << "Before:"; for (size_t i = 0; i < a.size(); ++i) { std::cout << " (" << a[i] << ", " << b[i] << ")"; } std::cout << std::endl; typedef std::vector<size_t>::iterator vec_iter; typedef boost::zip_iterator<boost::tuple<vec_iter, vec_iter> > zip_iter; std::sort(zip_iter(boost::make_tuple(a.begin(), b.begin())), zip_iter(boost::make_tuple(a.end(), b.end())), compare_first_elements_in_tuples()); std::cout << "After:"; for (size_t i = 0; i < a.size(); ++i) { std::cout << " (" << a[i] << ", " << b[i] << ")"; } std::cout << std::endl; return 0; } On gcc 4.0.1, the sort operation results in the element (2, 1) (a pair of corresponding elements in a and b) being duplicated, while the element (1, 5) does not appear in the result at all: Before: (0, 1) (2, 1) (1, 5) (2, 3) (4, 7) After: (0, 1) (2, 1) (2, 1) (2, 3) (4, 7) I suspect the problem is related to the fact that zip_iterator does not return a "real" reference when dereferenced. Am I mistaken about this? Is the code above supposed to work? If not, is there an alternative way to sort one sequence while swapping elements in another in a corresponding way? I am not able to join the two input sequences into a single sequence of pairs because of other factors in the use case for this code. Thank you for your help. -- Jeremiah Willcock

1 0

General C++: Class Method Signatures Miss-Match
by Etienne Philip Pretorius 02 May '09

02 May '09

Hello list, I know this is not boost related, but I will appreciate it if anyone could shed some light on what is happening here? Compiler error: xml.cpp: In member function ‘xml::document& xml::document::parse(std::vector<unsigned char, std::allocator<unsigned char> >&)’: xml.cpp:25: error: no matching function for call to ‘xml::document::characters(std::basic_string<unsigned int, std::char_traits<unsigned int>, std::allocator<unsigned int> >)’ xml.hpp:24: note: candidates are: virtual void xml::document::characters(unicode::string) unicode::string definition starts: namespace unicode { typedef std::size_t code_point; class string : public std::basic_string<code_point> { Is "std::basic_string<unsigned int, std::char_traits<unsigned int>, std::allocator<unsigned int> >" not the same as "unicode::string"? If not why is it not? Etienne

7 15

[BGL] unordered_set<std::vector<edge_descriptor> >
by Johan Oudinet 02 May '09

02 May '09

Hi, I want to keep a set of paths, which are implemented as a vector of edge_descriptor. I don't care about the order. I just have to add paths and be able to know the number of different paths present in the set. So, I plan to use unordered_set but I'd like to know you opinion about a hash function, since AFAIK, unordered_set is like an hash_set? My main issue is to use as few memory as possible, because there will have a lot of paths in the set. Any idea? Thanks for your help. -- Johan

3 5

performance of dijkstra_shortest_paths
by Bhaskara Marthi 01 May '09

01 May '09

I recently ran into a situation where dijkstra_shortest_paths in the boost graph library was running very slowly. When I replaced the call to dijkstra_shortest_paths with the code below, things were a couple of orders magnitude faster. Any idea why? - Bhaskara Version 1: with dijkstra_shortest_paths ReachableCostVector GridGraph::singleSourceCosts (const Cell2D& source, const vector<Cell2D>& dests) { GridDistances distances; GridGraphVertex start = cellVertex(source); resetIndices(); typedef map<GridGraphVertex, GridGraphVertex> GridPreds; GridPreds predecessors; boost::dijkstra_shortest_paths (graph_, start, weight_map(get(&EdgeCost::cost, graph_)). vertex_index_map(get(&CellInfo::index, graph_)). distance_map(associative_property_map<GridDistances>(distances)). predecessor_map(associative_property_map<GridPreds>(predecessors)). // to avoid warnings visitor(GraphSearchVisitor())); ReachableCostVector v(dests.size()); transform (dests.begin(), dests.end(), v.begin(), bind(extractCost, this, distances, _1)); return v; } Version 2: without dijkstra_shortest_paths // Approximation to Dijkstra ReachableCostVector GridGraph::singleSourceCosts (const Cell2D& source, const vector<Cell2D>& dests) { GridDistances distances; GridGraphVertex start = cellVertex(source); typedef pair<GridGraphVertex, double> QueueItem; queue<QueueItem> q; set<GridGraphVertex> visited; q.push(QueueItem(start,0.0)); visited.insert(start); while (!q.empty()) { GridGraphVertex v; double d; tie(v,d) = q.front(); q.pop(); GridGraphAdjacencyIterator iter, end; for (tie(iter,end)=adjacent_vertices(v, graph_); iter!=end; iter++) { if (visited.find(*iter)==visited.end()) { double cost = d+graph_[edge(v,*iter,graph_).first].cost; q.push(QueueItem(*iter, cost)); visited.insert(*iter); distances[*iter] = cost; } } } ReachableCostVector costs; for (vector<Cell2D>::const_iterator iter=dests.begin(); iter!=dests.end(); ++iter) { GridDistances::iterator pos = distances.find(cellVertex(*iter)); if (pos==distances.end()) costs.push_back(ReachableCost(false,-42.0)); else costs.push_back(ReachableCost(true, pos->second)); } return costs; }

4 7

New declaration ambiguates old declaration mutant.hpp
by Celine.Murphy＠barclayscapital.com 01 May '09

01 May '09

/ThirdParty/include/boost-1.38.0/boost-1_38/boost/bimap/relation/detail/ mutant.hpp:68: new declaration `template<class View, class Type> boost::enable_if<boost::is_const<Reference>, const View&>::type boost::bimaps::relation::detail::mutate(Type&)' /ThirdParty/include/boost-1.38.0/boost-1_38/boost/bimap/relation/detail/ mutant.hpp:55: ambiguates old declaration `template<class View, class Type> boost::enable_if<boost::mpl::not_<boost::is_const<Reference> >, View&>::type boost::bimaps::relation::detail::mutate(Type&)' Hi, I am getting the above error when I try to compile a class that includes boost/bimap.hpp. It happens at the include stage. Has anyone seen this before? Gcc 3.2.3, boost 1.38. Thanks Celine _______________________________________________ This e-mail may contain information that is confidential, privileged or otherwise protected from disclosure. If you are not an intended recipient of this e-mail, do not duplicate or redistribute it by any means. Please delete it and any attachments and notify the sender that you have received it in error. Unless specifically indicated, this e-mail is not an offer to buy or sell or a solicitation to buy or sell any securities, investment products or other financial product or service, an official confirmation of any transaction, or an official statement of Barclays. Any views or opinions presented are solely those of the author and do not necessarily represent those of Barclays. This e-mail is subject to terms available at the following link: www.barcap.com/emaildisclaimer. By messaging with Barclays you consent to the foregoing. Barclays Capital is the investment banking division of Barclays Bank PLC, a company registered in England (number 1026167) with its registered office at 1 Churchill Place, London, E14 5HP. This email may relate to or be sent from other members of the Barclays Group. _______________________________________________

2 2

boost::posix_time::from_time_t bug ???
by Łukasz Majewski 01 May '09

01 May '09

Hello, i think that i've found bug in date time library or maybe i made something wrog. SAMPLE CODE: #include <boost/date_time/posix_time/posix_time.hpp> #include <iostream> using namespace boost::posix_time; int main( int argc, char** argv ) { time_t t = time( 0 ); std::cout << "time_t: " << ctime( &t ) << std::endl; struct tm* t2 = localtime( &t ); std::cout << "struct tm: " << asctime( t2 ) << std::endl; ptime t3 = from_time_t( t ); std::cout << "ptime from time_t: " << to_simple_string( t3 ) << std::endl; ptime t4 = ptime_from_tm( *t2 ); std::cout << "ptime from tm: " << to_simple_string( t4 ) << std::endl; return 0; } OUTPUT: time_t: Fri May 1 14:22:01 2009 struct tm: Fri May 1 14:22:01 2009 ptime from time_t: 2009-May-01 12:22:01 // incorrect result ptime from tm: 2009-May-01 14:22:01

2 2

[boost-users][StateChart] polymorphic_downcast failure
by Igor R 01 May '09

01 May '09

Hi, The above failure occurs in reaction_dispatcher::derived_with_action::react(). IIUC, state_machine tries to invoke in_state_reaction that's defined for an event, which is unrelated to the one being processed. What logical error in my state-machine might cause such a problem? Unfortunately, I don't have any self-contained code at the moment, so please treat my question as theoretical :). Thanks.

2 6

[boost-users][python]
by Germán Diago 01 May '09

01 May '09

I have a question about managing python objects. I have a base class implemented in c++, which I use in python. >From python I inherit that class. And the I try to use the python instance from c++, like this. //This class has a function which is registered in python. I want to //be able to register python instances with this class class AnotherClass { void registerobject(const std::string & name, MyBaseClass * a) { v.push_back(a); } }; //Interface class, which I inherit in python class MyBaseClass { virtual const std::string getData(Object * o) const = 0; }; >From python: class MyPyClass(MyBaseClass): def __init__(self): MyBaseClass.__init__(self) def getData(self, o): return "hello" My question is. In order to register that class, namely, MyPyClass(which is a python class): can I do it through registerobject? Or should I implement something like this?: class AnotherClass { void registerobject(const std::string & name, python::object & o) { MyBaseClass * m = extract<MyBaseClass>(o); v.push_back(m); } }; In all two cases I get error_already_set exception when I try to execute the registerobject function and don't know why.

1 1

[statechart] Asynch statechart in dll
by Bill Clark 01 May '09

01 May '09

I have modified the Pingpong sample (an asynchronous statechart) so that the statechart itself is in a dll, with the calling code (in PingPong.cpp) in an exe. This is using VS2005 sp1. I have used a macro in the usual way to mark the Player class as exported. The app builds and runs, but it seems that events don't get acted on. We hit Player's initiate_impl(), and we get to state_machine::process_event() with a BallReturned event passed in, but we never hit the handler for that event (Waiting::react(const BallReturned&)). Is this something that should work? Is it necessary to do something more than just export the necessary classes from the dll? Thanks, Bill **

2 4

Overloading errors
by A B 01 May '09

01 May '09

Does anyone know why the following sample fails (it was adapted from the boost overloading example) and how to do this correctly. Thanks in advance. #include <boost/python/module.hpp> #include <boost/python/def.hpp> #include <boost/python/args.hpp> #include <boost/python/tuple.hpp> #include <boost/python/class.hpp> #include <boost/python/overloads.hpp> #include <boost/python/return_internal_reference.hpp> using namespace boost::python; struct X { int f(int x, double y = 4.25, char const* z = "wow") { return x; } int f(int x, bool b = true) { return x; } }; BOOST_PYTHON_MEMBER_FUNCTION_OVERLOADS(f_member_overloads, X::f, 1, 3) BOOST_PYTHON_MODULE(args_ext) { class_<X>("X", "This is X's docstring") .def("f1", &X::f, f_member_overloads(args("x", "y", "z"), "f's docstring" )[return_internal_reference<>()] ) ; } g++ -fpic -shared -o test.so test.cpp -lpython2.4 -lboost_python -I/usr/include/python2.4 test.cpp: In function ‘void init_module_args_ext()’: test.cpp:31: error: no matching function for call to ‘boost::python::class_<X, boost::python::detail::not_specified, boost::python::detail::not_specified, boost::python::detail::not_specified>::def(const char [3], <unresolved overloaded function type>, boost::python::detail::overloads_proxy<boost::python::return_internal_reference<1ul, boost::python::default_call_policies>, f_member_overloads>)’

2 2

Boost.Date_Time building for Windows CE
by Tory Bjorklund 01 May '09

01 May '09

Hello, My application builds fine when using Boost.Date_Time with win32 but when building with a configuration for Windows CE I get the following errors: boost/date_time/c_time.hpp(83) : error C2039: 'localtime' : is not a member of 'std' boost/date_time/c_time.hpp(83) : error C2660: 'boost::date_time::c_time::localtime' : function does not take 1 arguments boost/date_time/c_time.hpp(92) : error C2039: 'gmtime' : is not a member of 'std' boost/date_time/c_time.hpp(92) : error C2660: 'boost::date_time::c_time::gmtime' : function does not take 1 arguments boost/date_time/gregorian/gregorian_io.hpp(12) : fatal error C1083: Cannot open include file: 'locale': No such file or directory Does anyone know how I may be able to build an app for CE that includes the Date_Time library? Thanks, Tory

1 0

Boost.MultiIndex question on merge
by arm2arm 01 May '09

01 May '09

Hello, Are there way to merge two multiindex objects ? suppouse we have : struct strData{ int id; float a, b; strData(int id_, float a_, float b_):a(a_),b(b_),id(id_){} }; typedef multi_index_container< strData, indexed_by< ordered_unique< tag<id>, BOOST_MULTI_INDEX_MEMBER(strData,int,id)>, ordered_non_unique< tag<snap>,BOOST_MULTI_INDEX_MEMBER(strData,float,a)> > > data_setA, data_setB; Are there way to do: data_setA.merge(data_setB) ? thank you beforehand. Arman. -- View this message in context: http://www.nabble.com/Boost.MultiIndex-question-on-merge-tp23277496p2327749… Sent from the Boost - Users mailing list archive at Nabble.com.

5 11

boost coroutine status
by Robert Ramey 30 Apr '09

30 Apr '09

This package has been in the vault or some time. At least one well known person on this list has indicated that the quality is good and that he is using it in production code. Turns out that I now have a need of this facility but I've got a couple of problems. a) When I try and build it, I get ConvertFiberToThread which looks to me that there is some bjam magic missing. b) I would like to look the documentation. But that's hard to do as it's in quick book files and I don't have a quick book reader. I don't know that I have the ability to generate html files from quickbook. Maybe I do, but I can't really figure this out. c) I don't think that this package has ever been reviews nor do I see it scheduled for review. What is the status of this library. Anyone having information on this topic is encouraged to bring me up to date. Robert Ramey

5 13

[lexical_cast] float-to-string rounding error (not found with cout)
by Grant Birchmeier 30 Apr '09

30 Apr '09

I'm seeing unexpected behavior when using lexical_cast<string>. This code: const float f = 00123.12300; cout << f << endl; cout << boost::lexical_cast<std::string>(f) << endl; gives the result: 123.123 123.123001 Two questions: (1) Is there a way to eliminate this trailing '1' with lexical_cast (presumably some wierd rounding error)? (2) When casting float-to-string, given this behavior, why should I use lexical_cast instead of, say, stringstream? Thanks -Grant

4 5

Default argument values with boost::function
by Artyom 30 Apr '09

30 Apr '09

Hi, Is there any way to set default argument values with Boost::Function? On VS2008 SP1 I tried: boost::function< void( unsigned i = 1 ) > increment; But this didn't work. Code such as increment( 1 ); compiles OK but if you try doing increment(); the following compiler error is triggered: error C2064: term does not evaluate to a function taking 0 arguments class does not define an 'operator()' or a user defined conversion operator to a pointer-to-function or reference-to-function that takes appropriate number of arguments Thanks, Artyom

5 4

[boost-users][python] Pointer argument failed conversion?
by Germán Diago 30 Apr '09

30 Apr '09

Hello. After much googling I didn't find anything useful. My problem is: I have a virtual function like this class MyClass { void monitor(Object * o); } I have wrapped another class, which inherits from Object: class Character : public Object {...}; which is also exposed to python. I register my virtual function as the documentation says I have to do: class_<MyClass, boost::noncopyable>("MyClass"). def("monitor", &MyClass::monitor); When I try to call it from python, I get the following error: a = MyClass() b = Character(....) a.monitor(b) MyClass.monitor(MyClass, Character) did not match C++ signature: monitorizar(MyClass {lvalue}, Object *) Shouldn't my Character be automatically converted to Object * and then to make a successful call to the function? Should I do anything special to use pointers as parameters of my functions? I don't know why it tries to pass a "Character". What I want is to get it converted to Object *. Thanks in advance.

2 1

Program options required arguments?
by David Doria 30 Apr '09

30 Apr '09

If I have a list of 10 parameters, and all are required, I end up with a giant block of these: if(!vm.count("input")) { std::cout << "Input was not set." << std::endl; exit(-1); } Is there a way to do something like: desc.add_options() ("input", po::value<std::string>(&InputFile)->required(), "Set input file.") That will perform that check for me? Thanks, David

2 1

[interprocess]message_queue multi-client
by Laurence Chambenois 30 Apr '09

30 Apr '09

Hi, I need to communicate from x clients processes to one server process. I want to use the message_queue to send messages from the clients to the server, and when the server is done, it sends back its answer to the client. Do you have any examples ? with one bidirectionnal message_queue, or do I have to create a message_queue for each "direction" ? How can I deal with one message_queue and multiple process so that the request sent to one process is read by the good process ? Thanks for your help Laurence

1 0

Boost (1.38) on Intel Compiler (11.0)
by Jared White 30 Apr '09

30 Apr '09

Hey there, Was curious if someone can answer for me, does Boost support the intel compiler? I seem to be having some trouble linking to the libraries that are created when I "bjam" on Mac OSX using the Intel compiler. It builds the libraries as you would expect, and a whole bunch of them get dropped into the "lib" folder. In VC++, I give the linker this path, and it finds the libs, just fine. However, When I compile my application (with XCode), it can't find the libs. It would appear that the <boost/auto_link.hpp> is configuring the lib name to have a toolset of "iw" and not "il", as expected for Mac OS X. Any help is appreciated, -Jared

2 1

German Boost book updated
by Boris Schaeling 30 Apr '09

30 Apr '09

For those who know German: I've added five more chapters to my Boost book at http://www.highscore.de/cpp/boost/. There are now altogether 16 chapters covering 29 Boost libraries. The book does not try to replace the documentation but is written in a tutorial style to give developers a kick start. It explains what the numerous libraries are good for and how they are used without going too much into detail. Any feedback is, as always, appreciated! Boris

3 2

Multi Index: Nested std::pair
by Etienne Philip Pretorius 30 Apr '09

30 Apr '09

Hello List, I would greatly appreciate if someone could assist me in getting this structure correct. boost::multi_index_container< std::list< std::pair< /*value*/ unsigned char, std::pair< /*x co-ordinate*/ unsigned char, /*y co-ordinate*/ unsigned char > > >, boost::multi_index::indexed_by< /*value*/ boost::multi_index::ordered_non_unique< boost::multi_index::identity<unsigned char> >, /*x co-ordinate*/ boost::multi_index::ordered_non_unique< boost::multi_index::identity<unsigned char> >, /*y co-ordinate*/ boost::multi_index::ordered_non_unique< boost::multi_index::identity<unsigned char> > > > m; I do not see how "boost::multi_index::ordered_non_unique" will index the appropriate field in the nested std::pair structures. This is why I suspect there to be a problem with this structure. So before I continue, could someone please tell me if I am on the correct path. Grateful, Etienne

7 24

[Serialization] Getting crash in serialization library specific to Linux and 1.37
by Reusser, Edward 29 Apr '09

29 Apr '09

I created several DLL's that save and restore data using the serialization library. These libraries can be used on both the Windows and Linux and have been quite reliable since they were first created using the 1.34 boost release. The purpose of these libraries is to read and write serialized versions of our projects. We are using both XML and Text archives. The problem I am now looking at is in reading a Text archive. We recently upgraded to boost release 1.37. On Windows we are still using the 7.1 compiler, while on Linux we are using the gcc 3.2 compiler. We had previoously upgraded compilers without any problems, so its not a compiler upgrade issue. Upon upgrading to boost release 1.37, everything has continued to work just fine on Windows. However in the Linux builds, any designs written with the older 1.34 boost release now cause segmentation faults when being read with the more recent build. I have attached a stack trace. To complicate this picture, I can take a project written using the older 1.34 release on Linux, read it in Windows, then write it using Windows, then read it just fine in the new Linux build. Any ideas? Edward Reusser Principal Engineer Actel Corporation (650) 318-4972 The information contained in or attached to this email may be subject to the Export Administration Regulations (EAR), administered by the U.S. Department of Commerce, or the International Traffic in Arms Regulations (ITAR), administered by the U.S. Department of State, and may require an export license from the Commerce or State Department prior to its export. An export can include a release or disclosure to a foreign national inside or outside the United States. Include this notice with any reproduced portion of this information.

1 0

Cannot find -lboost_thhread with upgrade from 1.35 to 1.37
by Johan Mazel 29 Apr '09

29 Apr '09

Hi I just upgraded my OS from Ubuntu 8.10 to 9.04 and Boost upgraded at the same time from 1.35 to 1.37. I get some errors linked to the compilation command with g++ : /usr/bin/ld: cannot find -lboost_thread /usr/bin/ld: cannot find -lboost_filesystem These compilation options were working with the 1.35 version of Boost. I'd like to know wether there has been a change in Boost concerning the compilation options or if anything else have changed that would cause that problem. Thanks in advance for the answers. Johan Mazel

4 6