- Boost-users - lists.boost.org

[MSM] How to access event-attributes through a "kleene"-event?
by Deniz Bahadir 24 Feb '13

24 Feb '13

Hello everyone, I am trying to use Boost.MSM with the eUML front-end and I must say: That is a pretty cool thing! :-) However, I stumbled across some points that raised several questions, which I hoped someone of you could answer. One of these points is the usage of the "kleene" event (with eUML). The questions that came to mind are the following: 1. Is there a way to access an attribute of an (original) event if the transition-table uses the "kleene" event? 2. If there is a way, can I access it directly from the guard or actions-list in the transition-table? 3. If there is not a way, is it planned to be implemented (if possible) and when? For better understanding I insert here a simple example which explains what I am trying to do: In my special case, all my events have an attribute "EventNo" which is a unique number for each event(-type). Now, I need a transition for the events with "EventNo" greater than a special number (e.g. 7). So I thought, I could do something like this: // The attribute: BOOST_MSM_EUML_DECLARE_ATTRIBUTE(int, EventNo) BOOST_MSM_EUML_ATTRIBUTES((attributes_ << EventNo), EventNoAttr) // All events that contain this attribute: BOOST_MSM_EUML_EVENT_WITH_ATTRIBUTES(event1, EventNoAttr) BOOST_MSM_EUML_EVENT_WITH_ATTRIBUTES(event2, EventNoAttr) /* ... */ BOOST_MSM_EUML_EVENT_WITH_ATTRIBUTES(eventN, EventNoAttr) // Entry-action of FSM which initializes these event-numbers: BOOST_MSM_EUML_ACTION(init_all_events) { template <class Evt, class Fsm, class State> void operator()(Evt const&, Fsm&, State&) { event1.get_attribute(EventNo) = 1; event2.get_attribute(EventNo) = 2; /* ... */ eventN.get_attribute(EventNo) = N; }; } // Some states: BOOST_MSM_EUML_STATE((), State1) BOOST_MSM_EUML_STATE((), State2) // The transition table which contains all the logic: BOOST_MSM_EUML_TRANSITION_TABLE(( State1 + kleene [event_(EventNo) > Int_<7>()] // <-- Not working! == State2, State1 + event1 == State3 /* event2 to event7 are ignored! */ ), stt) // The FSM front-end and back-end: BOOST_MSM_EUML_DECLARE_STATE_MACHINE(( stt, init_ << init_all_events ), FsmFrontend; typedef msm::back::state_machine<FsmFrontend> Fsm; However, the expression "event_(EventNo)" in the transition-guard fails to compile with the following error: error: no match for call to ‘(boost::msm::front::euml::GetEvent_<EventNo_>) (const boost::msm::front::euml::kleene_&, boost::msm::back::state_machine< boost::msm::front::euml::func_state_machine<FsmFrontendtag, ...’ I read in the documentation of the Functor front-end that I should use boost::any::type() to access the original event in transition actions, but I do not really know how to go on with it. I am also not sure if this applies to the eUML front-end at all (especially in the transition-guard). I hope someone can help me and answer these questions!? Thanks in advance, Deniz Bahadir

2 6

[asio] Adding methods directly to the IO_Service for it to call.
by Ryan 24 Feb '13

24 Feb '13

Boost Community, I've created a basic class that wrapped the boost asio framework receive/send functionality. My wrapper class gave the packet buffer to whatever processing function that it was created with and when the function was done sent any buffers it was given. This let me create classes that processed data without any knowledge of the transport layer. This concept has worked well since the asio sockets have an async_receive and async_send. I was able to setup all of my classes and then call io_service.run() to get everything going. I now have to receive data from a customer dll and they only support polling. Is there a way to add to the io_service a specific method to call once the io_service is running. I've included the ReceiveProcessSend.hpp with a comment above the two async_receive calls I need to replace. Also, below is a very basic example of how I setup classes that I pass into the ReceiveProcessSend. Any help on how to add calling methods to the io_service would be really appreciated. Ryan class DoSomeProcessing { private: typedef std::vector<unsigned char> t_ucvector; typedef boost::shared_ptr<t_ucvector> t_boostVector; typedef std::list<t_boostVector> t_boostList; public: t_boostList const processMessage(t_boostVector message, t_ucvector::size_type messageSize) { t_boostList outputList; //Do processing here and create output buffer. return outputList; } }; int main(int argc, char * argv[]) { DoSomeProcessing myProcess; ReceiveProcessSend::Arguments myArguments; boost::asio::io_service io_service; ReceiveProcessSend rps( io_service, boost::bind(&DoSomeProcessing::processMessage, &myProcess, _1, _2), myArguments); io_service.run(); return 0; }

1 0

Comparing two mpl::vector types
by Rodrigo Madera 23 Feb '13

23 Feb '13

Hello all, Can anyone give me a hint on how to call a functor on the differences of two same size mpl::vector types? For example, given: typedef boost::mpl::vector<int, char, short, double> v0_type; typedef boost::mpl::vector<int, int, short, std::string> v1_type; We can see that the second and third elements don't match, so I wish to call a functor twice, with the following signature: struct SomeFunctor { template <typename T, typename U> void operator()(T t, U u) const { // do something nice here } }; In this example, functor would be called twice: once with [T=char, U=int] and another with [T=short, U=std::string], along with their values. Thanks for any light, Rodrigo

2 2

Geometry buffer algorithm: complete or not?
by John Lilley 23 Feb '13

23 Feb '13

I was looking here: http://www.boost.org/doc/libs/1_52_0/libs/geometry/doc/html/geometry/refere… And saw this: "The current implemetation only enlarges a box, which is not defined by OGC. A next version of the library will contain a more complete implementation" It looks like Boost::Geometry is incapable of buffering an arbitrary polygon. But various forum discussions refer to the buffer() algorithm like it actually works for the general case. Which is correct? John Lilley

2 8

Building / installing only the geometry library
by John Lilley 23 Feb '13

23 Feb '13

Is there a recommended approach to building/installing only the geometry library and its required headers? I understand that geometry is header-only, so I don't really need the libraries at all, but the build process that I am seeing to move the headers to the target folder builds them anyway. I am trying to get a minimal checkin to our version control that contains only those headers needed for geometry. Thanks, john

2 2

[LockFree] a queue of std::shared_ptr?
by Scott Smedley 22 Feb '13

22 Feb '13

Hi all, Is it possible to store std::shared_ptr objects in a boost::lockfree::queue? eg. boost::lockfree::queue<std::shared_ptr<Message>> q; gcc 4.7.2 with Boost 1.53.0 is giving a static assertion failure. It wants: boost::has_trivial_destructor<T>::value and boost::has_trivial_assign<T>::value Any help would be muchly appreciated. Scott.

4 4

scope_lock question
by Jeff Dunlap 22 Feb '13

22 Feb '13

I plan to start using Boost Thread, but for now I have a multithreaded Win32 application that does not use boost. Although the threads were not created with Boost Thread, can I use boost::mutex::scoped_lock to protect some new functions I write for the app? Example: boost::mutex mutex; void whatever() { boost::mutex::scoped_lock lock(mutex); // will this be protected even though the // thread was not created using Boost? } Thank you

3 2

[Building] Weird crash on windows while trying to build boost
by Alessandro Stamatto 21 Feb '13

21 Feb '13

Hi, I'm trying to build Boost on windows 7, I'm using rubenv mingw+gcc-4.8 (from here: http://sourceforge.net/projects/mingw-w64/files/Toolchains%20targetting%20W… ). I tried the usual way: "bootstrap gcc" and ".\b2". But at the beginning of the building process a weird crash happens with a windows message dialog of "Keyboard History Utility has stopped working". It happens even with the suppressing option ('-d0'). In the process list I see that this Keyboard Utility is called "doskey.exe". What is happening? Is it possible to build boost with gcc-4.8 on windows? Thanks for your attention, Alessandro Stamatto.

1 0

New gil::io has been added
by Christian Henning 21 Feb '13

21 Feb '13

Hi all, I have successfully replaced the old gil::io extension with the new gil::io. The new io has been developed on and off over the last few years and I believe it's a great improvement over the old one. Before going into details I like thank everyone that has been helping me out along the way. I have learned a lot of things and I'm very grateful for the experience! There are numerous improvements and perhaps it's best to study the documentation before using it. Here are some noteworthy points that come to my mind: * updating code that uses the old version only requires to change the header from "boost/extension/io/jpeg_io.hpp" to "boost/extension/io/jpeg/jpeg_io_old.hpp" * new formats: bmp, pnm, raw, targa * read numerous meta information from images. See boost/gil/extension/io/png_tags.hpp for example. * use all sorts of file metaphers or file names, like FILE*, ifstream, stringstream, wstring, filesystem::path, etc * in-memory images are supported * read large images with a scanline_read_iterator * have access to the file format backend * etc. There are some items left for which I need help from the boost maintainers. * How do I hook up the gil::io documentation so it's build when all boost documentation is build? * Same for the test suite. * Should I update boost_release? Or should the new extension "simmer" for a while? Again thanks for everything, Christian

2 1

time out for r_c_shortest_paths
by Brammert Ottens 21 Feb '13

21 Feb '13

Hi, I want to be able to put a time out on the r_c_shortest_paths, and still be able to get the optimal paths that have been found so far. Currently, I changed the REF such that it always returns false as soon as the time out has reach, in effect emptying the unprocessed_labels queue. The problem with this is that the algorithm still needs quite some time to actually empty the queue. I tried to have the on_label_popped(cur_label) throw an exception when the timeout has been reached. The timeout itself works like a charm, but because the pareto_optimal_solutions container is only filled after the while loop I can't access the already found optimal solutions. Does anybody have an idea on how to do this in a better way, without having to request an addition to the boost API? Regards, Brammert

2 1