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

boost variant + boost serialization unused type
by Merrill Cornish 09 Feb '15

09 Feb '15

Using boost::variant with boost::serialization gives me a warning about unused type in the boost::serialization namespace: D:/Boost/boost/serialization/variant.hpp:69:75: warning: typedef 'types' locally defined but not used [-Wunused-local-typedefs] typedef typename boost::variant<BOOST_VARIANT_ENUM_PARAMS(T)>::types types; I looked up boost::ignore_unused, which says it's for unused types also, but everything I've tried for using in my code (following the definition of my variant) results in a syntax error or other bug claim. How should I use ignore_unused in this case?

2 1

[Serialization] XML: float format is scientific instead of human-readable since Boost 1.57
by Frank Stähr 09 Feb '15

09 Feb '15

Hello Boost guys, when I switched from Boost 1.55 to 1.57 there was a change in the formation of floats and doubles for xml archives. Look at this example: #include <fstream> #include <boost/archive/xml_oarchive.hpp> class MyClass { friend class boost::serialization::access; template<class Archive> void serialize(Archive & ar, const unsigned int version) { ar & boost::serialization::make_nvp("accuracy", accuracy); } double accuracy = 0.03; }; int main() { MyClass obj; std::ofstream ofs("sample.xml"); boost::archive::xml_oarchive oa(ofs); oa & boost::serialization::make_nvp("myclass", obj); } I compiled it with VS 2013 for 32 bit. The output file has the line: <m_accuracy>2.99999999999999990e-002</m_accuracy> In Boost 1.55 it was much better human-readable: 0.03. Is this behaviour (this change) intended? In any case, what is the best way to format floats and doubles? (i. e. a manipulator for the ofstream?) Thanks, Frank -- Frank Stähr Technische Universität Berlin Communication Systems Group Sekr. EN1 Einsteinufer 17 10587 Berlin, Germany

5 10

write starvation of upgradable lock
by David Frank 09 Feb '15

09 Feb '15

From this discussion, http://stackoverflow.com/questions/21773618/starvation-with-upgrade-lock?no… There seems to be a starvation of writer with upgradable lock, I'm wondering if there is any fix/improvement with later release, e.g, 1.56, or 1.57 Thanks,David

2 1

[website] File appears truncated
by Joaquin M Lopez Munoz 07 Feb '15

07 Feb '15

The following file (which is linked from Boost.MultiIndex docs): http://www.boost.org/doc/libs/1_57_0/libs/multi_index/example/hashed.cpp appears truncated at line 8, yet it is perfectly OK in github: https://github.com/boostorg/multi_index/blob/master/example/hashed.cpp Any clue what's going on? Thank you, Joaquín M López Muñoz Telefónica

2 1

[System] Link issues using BOOST_ERROR_CODE_HEADER_ONLY
by Jonathan Jones 07 Feb '15

07 Feb '15

Hello, We're in the process of upgrading from Boost 1.49 to Boost 1.56. As part of this upgrade, we've noticed the linkage requirements for using the "system" library are much more strict in 1.56. We are getting linker errors like the following across dozens of our libraries: boost/system/error_code.hpp: undefined reference to `boost::system::generic_category()' boost/system/error_code.hpp: undefined reference to `boost::system::system_category()' We decided to fix these by defining the macro BOOST_ERROR_CODE_HEADER_ONLY according to this link: http://www.boost.org/doc/libs/1_56_0/libs/system/doc/reference.html#Header- error_code However, when building Boost itself with this macro defined, we're encountering several problems: 1) Some Boost libraries (chrono, filesystem, random) that use the "system" library no longer link, as if they are still expecting the library to be present. Note that some libraries, like thread, that use "system", link fine without the shared library being present. 2) We get multiple definitions of "boost::system::throws" Any thoughts on how to resolve these? Google search has not turned up any bugs that I can find. Sincerely, Jonathan Jones

2 1

[msm] multi thread access
by Sam Fisher 05 Feb '15

05 Feb '15

Hi Christophe, Did you consider adding build-in multi-thread support? (via policies maybe?) For example, I can guaranteed that all access to sub-fsm within its handlers (process_event()) will be synchronized. The problem is, sub-fsm "knows" about its parent and internally communicates with it (forwarding events etc). I can't control it from my side. typical scenario: an action from sub-fsm struct A_Init { template <typename Evt, typename FSM, typename SourceState, typename TargetState> void operator()(Evt&, FSM& fsm, SourceState&, TargetState&) const { std::thread([=]() mutable { std::lock_guard<std::mutex> l(access); fsm.process_event(some_event()); }).detach(); } }; locking proccess_event() in sub-fsm doesn't help if it alters the root.

2 3

Using io_service for synchronous tcp accept + custom tasks
by Aaron Levy 04 Feb '15

04 Feb '15

I've written code of the form: Code: io_service svc; io_service::work work(svc); thread t1( [&svc]() { svc.run(); } ); thread t2( [&svc]() { svc.run(); } ); thread t3( [&svc]() { svc.run(); } ); endpoint ep(ip::tcp::v4(), port); acceptor acceptor(svc, ep); while (true) { shared_ptr<socket> sock(new socket(svc)); acceptor.accept(*sock); svc.post( [sock]() { /* do stuff on sock here */ }); } Is this way of using io_service for accepting tcp connections and also as a thread pool for serving connected clients valid or could I hit some undefined behavior. -- Aaron Levy aaron.levy(a)yandex.com

3 5

joining boost ranges in loop
by Ernest Zaslavsky 04 Feb '15

04 Feb '15

Actually the question is how to achieve it? The use of boost::any_range is possible but undesirable (performance considerations) Sample code: typedef boost::multi_index_container< int, bmi::indexed_by<bmi::hashed_unique<bmi::tag<struct someTag>, bmi::identity<int>>>> Datum; class bar { public: auto foo() const; private: Datum m_someDataContainer; }; SomeType bar::foo() const { SomeType retVal; for (auto i = 0u; i < 10; ++i) { retVal = boost::range::join(retVal, m_someDataContainer.equal_range(i)); } return retVal; } Here we can observe several problems. The return type cannot be easily deduced. The join would create joined_range<joined_range<joined_range and so on of nested types. The above can be achieved by using any_range like typedef boost::any_range<int, boost::random_access_traversal_tag, int&, std::ptrdiff_t> VersionRange; but above will create nested joined_range of any_range which, when the number of iterations is big enough, has tremendous impact on performance. So, what can be done here?

1 0

boost serialization + boost variant + boost blank
by Merrill Cornish 04 Feb '15

04 Feb '15

I'm running Boost 1.57.0 using serialization, variant, and boost::blank with that variant. I found that when compiled, serialization complains that blank does not contain a serialize method, which is true. I made a copy of blank.hpp, called it blank.h, and added template< class Archive > void serialize(Archive& ar, const unsigned int version) {} to the body of struct blank. That made the compiler happy, but is it a fix in general?

4 6

[statechart] fifo_scheduler<>::operator() in state constructor
by Andrew C 04 Feb '15

04 Feb '15

Hi, is it allowed to call fifo_scheduler<>::operator() from state constructor? If yes, then let's assume "scheduler" and "processor" are objects of "fifo_scheduler<>" and "fifo_scheduler<>::processor_handle" types respectively in the some state constructor code below: State(my_context ctx): my_base(ctx) { boost::intrusive_ptr<Event> pEvent(new Event()); // If both lines [A] and [B] are uncommented // then "unconsumed_event" method is called twice. // If just line [A] is uncommented then "unconsumed_event" method is called. // If just line [B] is uncommented then certain "react" method is called. // scheduler.queue_event(processor, pEvent); // [A] // this->post_event(Event()); // [B] scheduler(); } Note constructor calls "scheduler()" at the end. Other questions: Is this code correct and runs expectedly as described in comments? Why "unconsumed_event" is called if use "queue_event" instead of "post_event"? Thanks, Andrew.

1 0