- Boost-users - lists.boost.org

accumulators: not copying doubles into the acc set
by MM 01 Aug '12

01 Aug '12

Hello, Having a list of doubles stored somewhere, is there a way not to copy the doubles into the accumulator set and just iterate somehow through the existing data in order to save memory. Imagine I have a std::vector<double> with 10million entries. Then pushing each double to the accumulator set object, then I'd imagine, at one point, the process would hold the 10 million doubles twice. Rds, MM

3 3

about how to install mpi of the boost1.49
by Kazuhide Ota 01 Aug '12

01 Aug '12

Dear all, Hello all, My name is Kazuhide Ota. I tryied to install boost1.49, but I got a problem. When I tried to install it on cygwin, I ran bootstrap.sh and bjam.exe, and I received the following message: <message> warning: skipping optional Message Passing Interface (MPI) library. note: to enable MPI support, add "using mpi ;" to user-config.jam. note: to suppress this message, pass "--without-mpi" to bjam. note: otherwise, you can safely ignore this message. MPI is necessary for me, so I added "using mpi : mpic++ ;" to user-config.jam in /boost_1_49_0/tools/build/v2, and I ran bjam.exe again. However, I received the same message. Did I edit the different file? If someone know hot to solve this problem, could you teach me? Best regards, Kazuide Ota

1 0

program_options save config example
by Neil Nelson 01 Aug '12

01 Aug '12

This post goes toward providing a config or log file output of the input to Boost program_options. This was a requested capability but apparently without further work. See the lines under 'config out:' at the end. http://boost.2283326.n4.nabble.com/program-options-save-configuration-td256… The following Boost any page provides methods to convert the any values upon iterating the vm map. http://www.boost.org/doc/libs/1_50_0/doc/html/any/s02.html // program_options_wrapper.hpp // Latest boost program_options: // See http://boost-sandbox.sourceforge.net/doc/html/program_options.html // See http://svn.boost.org/svn/boost/trunk/libs/program_options/example/ #ifndef PROGRAM_OPTIONS_WRAPPER_HPP #define PROGRAM_OPTIONS_WRAPPER_HPP #include <boost/any.hpp> #include <boost/program_options.hpp> #include <string> namespace po = boost::program_options; class program_options_wrapper { public: po::variables_map vm; po::options_description cmd_line; po::options_description config; po::options_description hidden; po::options_description version; po::options_description all; // Singleton static program_options_wrapper& instance() { static program_options_wrapper instance_; return instance_; } bool initialize(int argc, char **argv); private: std::string config_file; // Singleton helpers program_options_wrapper() {} program_options_wrapper(program_options_wrapper const& copy); program_options_wrapper& operator=(program_options_wrapper const& copy); bool is_empty(const boost::any & operand); bool is_int(const boost::any & operand); bool is_double(const boost::any & operand); bool is_char_ptr(const boost::any & operand); bool is_string(const boost::any & operand); // void get_typeid(const boost::any & operand); }; // class program_options_wrapper #endif // #define PROGRAM_OPTIONS_WRAPPER_HPP // program_options_wrapper.cpp #include <boost/any.hpp> #include <boost/program_options.hpp> #include <string> #include <vector> #include <fstream> #include "program_options_wrapper.hpp" namespace po = boost::program_options; bool program_options_wrapper::initialize(int argc, char **argv) { version.add_options() ("version: program_options.cpp", "1.0") ("version: program_options.hpp", "1.0") ; cmd_line.add_options() ("help,h", "produce help message") ("version,v", "list component versions") ("config,c", "configuration file name") ("report,r", "report run options") ("vector,x", boost::program_options::value<std::vector<std::string> >(), "test string vector") ("double,d", boost::program_options::value<double>(), "test double") ; all.add(cmd_line).add(config).add(hidden); // po::store(po::parse_command_line(argc, argv, cmd_line), vm); // po::notify(vm); po::parsed_options parsed = po::command_line_parser(argc, argv).options(cmd_line).allow_unregistered().run(); po::store(parsed, vm); // po::store(po::command_line_parser(argc, argv).options(cmd_line).allow_unregistered().run(), vm); po::notify(vm); if (vm.count("config")) { std::ifstream ifs(config_file.c_str()); if (!ifs) { std::cout << "can not open config file: " << config_file << std::endl; return false; } else { po::store(parse_config_file(ifs, config), vm); po::notify(vm); } } if (vm.count("help")) { std::cout << all; return false; } if (vm.count("version")) { std::cout << version; return false; } if (vm.count("report")) { std::cout << all; std::cout << version << std::endl; std::cout << "command-line:"; for(int j=0;j<argc;j++) std::cout << " " << argv[j]; std::cout << std::endl << std::endl; std::cout << "config out:" << std::endl; for (po::variables_map::iterator it=vm.begin(); it != vm.end(); it++ ) { if (is_empty(it->second.value())) std::cout << "# " << it->first << "=is_empty" << std::endl; if (vm[it->first].defaulted()) std::cout << "# " << it->first << "=is_defaulted" << std::endl; if (is_int(it->second.value())) std::cout << it->first << "=" << vm[it->first].as<int>() << std::endl; else if (is_double(it->second.value())) std::cout << it->first << "=" << vm[it->first].as<double>() << std::endl; else if (is_char_ptr(it->second.value())) std::cout << it->first << "=" << vm[it->first].as<const char *>() << std::endl; else if (is_string(it->second.value())) { std::cout << it->first << "="; std::string temp = vm[it->first].as<std::string>(); if (temp.size()) std::cout << temp << std::endl; else std::cout << "true" << std::endl; } else { // Assumes that the only remainder is vector<string> std::vector<std::string> poVector = vm[it->first].as<std::vector<std::string> >(); for (std::vector<std::string>::iterator vecIt=poVector.begin() ; vecIt != poVector.end(); vecIt++ ) std::cout << it->first << "=" << (*vecIt) << std::endl; } } std::vector<std::string> unrecognized_options = po::collect_unrecognized(parsed.options, po::include_positional); if (unrecognized_options.size()) { std::cout << std::endl << "unrecognized options:"; for (int j=0;j<unrecognized_options.size();j++) std::cout << " " << unrecognized_options[j]; std::cout << std::endl; } std::cout << std::endl; } return true; } bool program_options_wrapper::is_empty(const boost::any & operand) { return operand.empty(); } bool program_options_wrapper::is_int(const boost::any & operand) { return operand.type() == typeid(int); } bool program_options_wrapper::is_double(const boost::any & operand) { return operand.type() == typeid(double); } bool program_options_wrapper::is_char_ptr(const boost::any & operand) { try { boost::any_cast<const char *>(operand); return true; } catch(const boost::bad_any_cast &) { return false; } } bool program_options_wrapper::is_string(const boost::any & operand) { return boost::any_cast<std::string>(&operand); } /* void program_options_wrapper::get_typeid(const boost::any & operand) { std::cout << operand.type().name() << std::endl; } */ // test.cpp /* program_options.o: program_options_wrapper.hpp program_options_wrapper.cpp g++-4.7 -g -c -Wall -std=c++11 program_options_wrapper.cpp -o \ program_options_wrapper.o test: test.cpp program_options_wrapper.o g++-4.7 -Wall -g -std=c++11 test.cpp -o test \ program_options_wrapper.o -lboost_program_options */ #include <boost/program_options.hpp> #include <iostream> #include "program_options_wrapper.hpp" int main(int argc, char **argv) { auto& po_wrap = program_options_wrapper::instance(); po_wrap.version.add_options()("version: build_state_tree.cpp", "1.0"); if (!po_wrap.initialize(argc, argv)) exit(0); return 1; } // test output -h [ --help ] produce help message -v [ --version ] list component versions -c [ --config ] configuration file name -r [ --report ] report run options -x [ --vector ] arg test string vector -d [ --double ] arg test double --version: build_state_tree.cpp 1.0 --version: program_options.cpp 1.0 --version: program_options.hpp 1.0 command-line: ./build_state_tree -r -x str1 -x str2 -d 345.67 -z testing config out: double=345.67 report=true vector=str1 vector=str2 unrecognized options: -z testing Neil Nelson

1 0

[context / coroutine] linker errors
by Szymon Gatner 31 Jul '12

31 Jul '12

I grabbed coroutine lib from http://olk.bplaced.net/boost-coroutine.zip and I am getting linker errors with VC2008 SP1 when trying to compile simplest example from coroutine's documentation: Error 1 error LNK2019: unresolved external symbol _jump_fcontext referenced in function "public: int __thiscall boost::coro::detail::context_base<void __cdecl(void),class boost::ctx::stack_allocator,void,0>::native_start(void)" (?native_start@?$context_base@$$A6AXXZVstack_allocator@ctx@boost@@X$0A@@detail@coro@boost@@QAEHXZ) main_2.obj context Error 2 error LNK2019: unresolved external symbol _make_fcontext referenced in function "public: __thiscall boost::coro::detail::context_base<void __cdecl(void),class boost::ctx::stack_allocator,void,0>::context_base<void __cdecl(void),class boost::ctx::stack_allocator,void,0>(class boost::ctx::stack_allocator const &,unsigned int,enum boost::coro::flag_unwind_t,bool)" (??0?$context_base@$$A6AXXZVstack_allocator@ctx@boost@@X$0A@@detail@coro@boost@@QAE@ABVstack_allocator@ctx@3@IW4flag_unwind_t@23@_N@Z) main_2.obj context btw, old documentation link (http://ok73.ok.funpic.de/boost/libs/context/doc/html/) for boost.context does not work anymore and I remember it had continuation example which is not included in library examples. Any way to get it back? -- Szymon Gatner The Lordz Games Studio www.thelordzgamesstudio.com

2 1

Re: [Boost-users] Boost 1.49 msm And_ operator
by Kamel Zaarouri, Mr 30 Jul '12

30 Jul '12

Hi, I know this may seem silly but is there an operator that I can use to combine actions in the transition table to same way I can "and" guard conditions? if I have 2 guards, g1 and g2 in the transition table I can write under guard And_<g1,g2> what if I want to execute 2 actions upon a transition, obviously I can't have And_<a1,a2> since actions do not return boolean. Any idea? Thanks. ________________________________________ From: boost-users-bounces(a)lists.boost.org [boost-users-bounces(a)lists.boost.org] on behalf of Christophe Henry [christophe.j.henry(a)googlemail.com] Sent: Wednesday, July 11, 2012 4:17 PM To: boost-users(a)lists.boost.org Subject: Re: [Boost-users] Boost 1.49 msm problems with events > Hi, I am having problems with my transition table and events. Let me > explain the fake design that illustrates the problem: > > I have a state machine (myStateMachine) containing 2 states (s0 and s1) > and 1 substatemachine (subm1). > > the substatemachine subm1 contains an initial state 'sub0' and also s1 > (the same state as in myStateMachine). > > this is the main transition table: > > s0->s1 on event 'ES1' > s0->s2 on event 'ES2' > s0->subm1 on event 'ESUB' > > this is the submachine transition table: > > sub0->s1 on event ES1 > > now, assume that state s1 is using the event that triggered it to extract > some information i.e. > > struct s1 : public msm::front::state<> > { > template <class Event,class FSM> > void on_entry(Event const& evt,FSM& fsm) > { > evt.getEventData(); > } > } > so every event that could transition to s1 needs to implement > getEventData() method. > > ->this is normal! > > now my problem is that ESUB does NOT implement getEventData() but > apparently it should. And I don't get why. > > I am not using ESUB to transition to s1 but I am using ESUB to transition > to subm1 and subm1 contains s1 but I don't access it at that point. > > I hope this is clear. > > Thank you. Hi, this is an unfortunate limitation of msm (for composites), which I have on my list of stuff to solve asap. The problem is that though the event esub is not used to transition to s1, for the compiler it could. Whatever, it's my fault, plus I forgot it in the doc :( The solution is to help the compiler a bit by enabling on_entry with evt.getEventData() only for events having a special property, like your es1. For example: BOOST_MPL_HAS_XXX_TRAIT_DEF(get_event_data) // this event supports getEventData struct es1 { typedef int get_event_data; void getEventData(){...} }; Then use this in your state: struct s1 : public msm::front::state<> { template <class Event,class FSM> typename boost::enable_if<typename has_get_event_data<Event>::type,void>::type on_entry(Event const& evt,FSM& fsm) { evt.getEventData(); } // for events not supporting getEventData like esub template <class Event,class FSM> typename boost::disable_if<typename has_get_event_data<Event>::type,void>::type on_entry(Event const& ,FSM& ) { } }; HTH, Christophe _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

2 1

Can I copy/paste Boost source code and compile directly into my application?
by Fred 29 Jul '12

29 Jul '12

Dear Boosters, I would like to use the Boost/regex library in my application, but I don't want my users to have to install Boost. My preference would be to copy a minimal set of Boost .hpp and .cpp files into a sub-directory of my application and then compile and link the boost/regex stuff in the same way as the rest of my application. Note that the regex library is not a header-only library. So, is there a reasonably straight-forward and well-accepted procedure to do this? Apologies if this topic has already been covered, but I can't seem to get a hit. Many thanks, Fred

6 7

Boost Python: add_property and shared_ptr
by Raul Durand 29 Jul '12

29 Jul '12

Hi! I've been using the boost::python library for some time with few problems but now I got stuck. Fortunately I isolated the problem. In the class Domain below I defined a property called solver which returns a shared_ptr<Solver> instance: *namespace bp=boost::python;** **class Solver **{ **public: ** bool _b; ** void func() { _b = true; } **};** **typedef shared_ptr<Solver> Solver_ptr;** ** class Domain **{ **public: ** Solver_ptr solver; ** void set_solver(Solver_ptr s) { solver = s; } **};** **typedef shared_ptr<Domain> Domain_ptr;** ** BOOST_PYTHON_MODULE(test) {** * * bp::class_<Solver, Solver_ptr>("Solver")* * .def("func", &Solver::func)* * ;* * bp::class_<Domain, Domain_ptr>("Domain")* * .def("set_solver" , &Domain::set_solver)* * .add_property("solver" , &Domain::solver)* * ;* *}* The code compiles ok. The problem appears while running in Python: *from test import * **d = Domain() **s = Solver() **d.set_solver(s) # is it ok to do that? **d.solver.func() # this line does not work* With the following error message: *Traceback (most recent call last):* * File "test_test.py", line 5, in <module>* * d.solver.func()* *Boost.Python.ArgumentError: Python argument types in* * Solver.func(Solver)* *did not match C++ signature:* * func(Solver {lvalue})* I suspect that the problem is related with add_property returning a shared_ptr<Solver> aka Solver_ptr. It seems that add_property is returning a *non lvalue* Solver object. I tried applying return value policies with no luck. How can I get Domain::solver with type Solver_ptr in Python as a lvalue ? Thanks! Raul Durand

1 0

[Iterator] swap and iter_swap do not work with zip_iterator
by Benjamin Lindley 28 Jul '12

28 Jul '12

I wanted to use zip_iterator to concurrently sort two arrays by the values in one of them. It didn't work. Most of the values were over-written by one of the pairs. I figured out that there was a problem with the swapping, so I made this small test case. Compiled with GCC 4.6.1 on Ubuntu 11.10 #include <utility> #include <iostream> #include <boost/iterator/zip_iterator.hpp> int main() { std::vector<int> a = { 1, 2 }; std::vector<int> b = { 100, 200 }; auto first = boost::make_zip_iterator( boost::make_tuple(a.begin(), b.begin())); auto last = boost::make_zip_iterator( boost::make_tuple(a.begin() + 1, b.begin() + 1)); std::cout << first->get<0>() << ' ' << last->get<0>() << '\n'; std::cout << first->get<1>() << ' ' << last->get<1>() << "\n\n"; using std::swap; using std::iter_swap; // swap(*first, *last); // doesn't compile iter_swap(first, last); std::cout << first->get<0>() << ' ' << last->get<0>() << '\n'; std::cout << first->get<1>() << ' ' << last->get<1>() << "\n\n"; } Output: 1 2 100 200 2 2 200 200 Expected Output: 1 2 100 200 2 1 200 100 Is this a known problem?

3 3

linking against boost libraries using a binary compatible compiler
by Petros 27 Jul '12

27 Jul '12

Hi, I am working on win7 with vs2010. Occasionally I want to compile with the windows intel compiler (icl, 12.1), whih is to be syntactically and binary fully compatible with the vs2010 (cl10.0). However, due to the “magic” that binds the boost code with the libraries to link against ( btw, where does this “magic” happen?), when I am building with intel, the intel libs are searched for. Building boost with the intel compiler on windows can have its moments, as witnessed by the repeated posts on the web . I would be perfectly happy with linking msvc-compiled boost with the rest of my code compiled with the intel compiler. How can this be done ? Has anyone done it ? (Someone must have thought of this – also saves from two separate builds of the boost libraries ) TVMIA, Petros

1 0

[boost] [fusion] fusion::map doesn't define 'category'
by Steve Lorimer 27 Jul '12

27 Jul '12

I am attempting to create a fusion map which looks like: map< pair<field1, int>, pair<field2, int> > I am then attempting to find out if *field1 *exists in the map using *has_key, *but am getting a compile error: error: no type named ‘category’ in ‘struct boost::fusion::result_of::as_map<boost::fusion::vector2<boost::fusion::pair<field1, int>, boost::fusion::pair<field2, int> > > Can anyone offer insight on what I'm doing wrong? I have a simple example below: #include <iostream> #include <boost/fusion/container.hpp> #include <boost/fusion/sequence.hpp> #include <boost/fusion/mpl.hpp> #include <boost/fusion/include/has_key.hpp> #include <boost/mpl/transform.hpp> namespace fusion = boost::fusion; namespace mpl = boost::mpl; *// given a field, returns a fusion pair of <field, field::type>* template<class field> struct make_pair { typedef typename fusion::result_of::make_pair<field, typename field::type>::type type; }; *// given a sequence of fields, returns a fusion map which maps field -> field::type* template<class fields> struct make_map { typedef typename mpl::transform<fields, make_pair<mpl::_1>>::type pair_sequence; typedef typename fusion::result_of::as_map<pair_sequence> type; }; *// sets boolean value to true if field is in fields* template <class field, class... fields> struct contains_key { typedef typename boost::fusion::vector<fields...> vector; typedef typename make_map<vector>::type map; typedef typename fusion::result_of::has_key<map, field>::type result; static const bool value = result(); }; struct field1 { typedef int type; }; struct field2 { typedef int type; }; int main() { std::cout << contains_key<field1, field1, field2>::value << std::endl; return 0; }

2 2