July 2012 - Boost-users - lists.preview.boost.org

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

[boost] [fusion] Find the index of an item in a fusion map
by Steve 27 Jul '12

27 Jul '12

I have a fusion map in a class which I want to initialize with a variadic constructor template<typename... Args> ctor(Args... args) { ..... } I imagine I would create a fusion vector using make_vector and then iterate over the elements of the map and somehow set the value of each map pair to it's corresponding element in the vector. I think that if I could find the index position of a single element in the map, that I could then use for_each to iterate over the pairs, and use the item's index to retrieve the corresponding value from the vector. Can anyone help me with this? Perhaps there is an alternate better way to do this? TIA Steve

1 0

[Serialization and C++11] Using std::unique_ptr with boost::serialization
by berenger.morel＠neutralite.org 26 Jul '12

26 Jul '12

Hello. I am trying to serialize a node of a DP composite. This would probably be straightforward if I used raw pointers, but I am using std::unique_ptr. I am not sure it is possible, since I tried to find the word unique_ptr in my /usr/include/boost directory and found almost nothing (except interprocess and chrono which have been installed for other purposes). So, is there is a way to use serialization on unique_ptr, or is there is a reason to avoid doing this? Note: I have made some tries with using boost::serialization on my objects as if they were raw pointers and it have obviously failed, because expected type seems to be only raw pointer and unique_ptr is not compatible with it.

2 1

[BGL] inv_adjacency_iterator is missing in graph_traits.hpp
by Johan Oudinet 26 Jul '12

26 Jul '12

Hi, I have a bidirectional adjacency list and I want to use the inv_adjacent_vertices method However, the following code does not compile: typedef boost::graph_traits<Graph> Traits; typedef typename Traits::inv_adjacency_iterator inv_adjacency_iterator; inv_adjacency_iterator w, w_end; boost::tie (w, w_end) = inv_adjacent_vertices (v, g_); The problem is that inv_adjacency_iterator is not part of boost::graph_traits :-( Could you tell me how can I use the inv_adjacent_vertices method (i.e., how can I declare an inv_adjacent_iterator)? Thanks. Best regards, P.S.: I tried to use an adjacency_iterator instead but it does not work. -- Johan () ascii ribbon campaign - against html e-mail /\ www.asciiribbon.org - against proprietary attachments

2 2

Boost 1_50 does not get installed
by Panagiotis Foteinos 26 Jul '12

26 Jul '12

Greetings. I invoke ./b2 variant=release and everything seems to be fine. I can see that the thread library is created inside under $SOURCE/bin.v2/libs . Nevertheless, the command ./b2 install does not copy the libraries under /usr/local. It does nothing. All I see is many warnings similar to: **************************** skipped bin.v2/libs/thread/build/gcc-4.4.3/release/link-static for lack of bin.v2/libs/thread/build/gcc-4.4.3/release... **************************** Any ideas? Regards, Panagiotis Foteinos OS: Ubuntu, gcc 4.4.3, 64 bits

2 4