March 2009 - Boost-users - lists.boost.org

Nstate: dynamic_bitset generalized for bounded integer values?
by Hostile Fork 06 Mar '09

06 Mar '09

Hello boost-users! I've written a class in the same spirit as the dynamic_bitset. But instead of being restricted to inserting values that range [0..1], the packed values can range from [0..radix-1]. Nothing that fancy, but a few people on #boost IRC thought it looked potentially useful, so I decided to try announcing it here. The library is called "Nstate", and is released under the boost software license (of course!) http://hostilefork.com/nstate/ As an example, I point out that it can fit 20 "Tristates" (a.k.a. Nstate<3>) into 32 bits, as opposed to 20 bytes (if one used a packing approach with 2 bits per tristate, you would only get 16 of them out of 32 bits). I'm guessing this would be low on the radar in terms of getting vetted for inclusion in boost. But if anyone out there is looking for something *like* this, here's a start. I'd certainly appreciate any peer review or wisdom to help make it better, faster, smarter...boostier? So if you're bored and just like making free libraries have more "Quality" then please send feedback! Files are: http://github.com/hostilefork/nocycle/blob/master/Nstate.hpp http://github.com/hostilefork/nocycle/blob/master/Nstate.cpp Regards, Brian

1 0

Re: [Boost-users] [Units] implicit conversion of quantities
by Christopher Bruns 05 Mar '09

05 Mar '09

Steven Watanabe wrote: > In this particular case, I would suggest writing the function as a template > and casting. In general you can write a forwarding overload that casts as > appropriate. Thanks Steven for the advice. I consulted Matthias Schabel off list to help figure out how to get this working. In case others have similar needs, I am attaching my modified test program. /* =================================== */ #include <boost/units/systems/si.hpp> #include <boost/units/base_units/metric/angstrom.hpp> #include <iostream> namespace boost { namespace units { typedef metric::angstrom_base_unit::unit_type angstrom_unit; BOOST_UNITS_STATIC_CONSTANT(angstrom, angstrom_unit); typedef scaled_base_unit< si::meter_base_unit, scale<10, static_rational<-9> > > nanometer_base_unit; typedef nanometer_base_unit::unit_type nanometer_unit; BOOST_UNITS_STATIC_CONSTANT(nanometer, nanometer_unit); } } // namespace boost::units using namespace boost::units; using namespace std; // First version of showBondLength takes a specific quantity type (nanometers) std::ostream& showBondLength(quantity<nanometer_unit> len, std::ostream& os) { // This implementation could be compiled into a library. os << len << std::endl; } // Second version of showBondLength can take any length type, and casts it to // the specific type. The parameter type is a bit long winded for a public API // method, but not unreadable. template<class Y> std::ostream& showBondLength(quantity<unit<length_dimension, Y> > len, std::ostream& os) { // This (simple, forwarding) implementation must be fully defined in a header. return showBondLength( (quantity<nanometer_unit>)len, os); } int main() { // These constructors all work; conversions are automatic quantity<nanometer_unit> len1(0.152 * nanometer); quantity<angstrom_unit> len2(1.52 * angstrom); quantity<nanometer_unit> len3(1.52 * angstrom); quantity<angstrom_unit> len4(0.152 * nanometer); // These three calls are OK, nanometer quantity is passed showBondLength(0.152 * nanometer, cout); showBondLength(len1, cout); showBondLength(len3, cout); // Explicit casts to nanometer are OK for angstrom quantities... // but yuck showBondLength( (quantity<nanometer_unit>)(1.52 * angstrom), cout ); showBondLength( (quantity<nanometer_unit>)len2, cout ); showBondLength( (quantity<nanometer_unit>)len4, cout ); // These all (used to) fail to compile before templated version was added. // Users want implicit conversion here from angstroms to nanometers showBondLength(1.52 * angstrom, cout); showBondLength(len1, cout); showBondLength(len2, cout); showBondLength(len3, cout); showBondLength(len4, cout); // Conversion from meters works too showBondLength(3.0 * si::meters, std::cout); // Non-length quantities correctly fail to compile // showBondLength(3.0 * si::seconds, std::cout); return 0; } /* =================================== */

1 0

[boost] compose ArgumentPacks using a comma
by er 05 Mar '09

05 Mar '09

Hi All, I have a structure containing a fusion sequence which is constructed recursively. BOOST_PARAMETER_KEYWORD(tag,kwd_struct) // This element depends on others, hence fetches the structure elem_x(const Args& args){ ...args[kwd_struct]...; } structure(Args const& args) :elements_( make_elements( ((kwd_struct = (*this)),args) ) ) {}; error: in constructor elem_x(const Args&) [with Args = ...] no match for operator[] in args[kwd_struct] If I reverse the order of the arguments (args,(kwd_struct = (*this))) it compiles. Why the discrepancy?

1 1

[accumulator] extract_result during initialization
by er 05 Mar '09

05 Mar '09

Hi All, I have an accumulator (a) that takes an initial value (x_), and another (b), whose initial value (x_) depends on that of (a). I would have thought that this is feasible because accumulator_set initializes (a) before (b), but apparently b.x_ is initialized with a garbage value. namespace impl { template<typename Sample,typename Id> class accu_a_impl : public accumulator_base { public: typedef Sample result_type; template<typename Args> accu_a_impl(Args const &args) :x_( args[kwd<Id>::value] ) { } result_type result(dont_care) const { return x_; } private: result_type x_; }; } namespace tag { template <typename Id = mpl::void_> struct accu_a : depends_on<> { typedef accumulators::impl::accu_a_impl<mpl::_1,Id> impl; }; } namespace impl { template<typename Sample,typename Id0,typename Id1> class accu_b_impl : public accumulator_base { typedef tag::accu_a<Id1> a_type; public: typedef Sample result_type; template<typename Args> accu_b_impl(Args const &args) :x_( extract_result<a_type>(args[accumulator]) // faulty? ) { } template<typename Args> void operator ()(Args const &args) { // overrides initialization for the sake of testing x_ = extract_result<a_type>(args[accumulator]); } private: result_type x_; }; } namespace tag { template <typename Id0,typename Id1=mpl::void_> struct accu_b : depends_on<tag::accu_a<Id1> > { typedef accumulators::impl::accu_b_impl<mpl::_1,Id0,Id1> impl; }; } *.cpp typedef mpl::size_t<0> id0; typedef mpl::size_t<1> id1; typedef double value_type; typedef tag::accu_a<id0> a_type; typedef tag::accu_b<id1,id0> b_type; typedef accumulator_set< value_type, stats< b_type > > set_type; set_type set( ( kwd<id0>::value = 0.9 ) ); std::cout << "a.x_=" << extract_result<a_type>(set) << std::endl; std::cout << "b.x_=" << extract_result<b_type>(set) << std::endl; set(0); std::cout << "b.x_=" << extract_result<b_type>(set) << std::endl; //Output: //a.x_=0.9 //b.x_=0.1 // where 0.9 is expected //b.x_=0.9

2 3

[statechart] StopWatch aborts on context<StopWatch>()
by jeti 05 Mar '09

05 Mar '09

hi all im pretty sure i just dont understand what context<>() really does and how it works reading the docs/tutorial doesnt seem to help me so, can someone point me to a documentation for boost::statechart and how to get a Pointer/Reference to the currents state_machine from a simple_state? what do i have to keep in mind? just to give an example: if i add the following line in the boost/libs/statechart/examples/StopWatch/StopWatch.cpp file Running() : startTime_( std::time( 0 ) ) { context<StopWatch>(); // <<< added } it compiles and abort's if i try to start the stopwatch i dont understand why? the state_machine is initialized and even the docs say in its comments that i can use it to get the statemachine http://www.boost.org/doc/libs/1_38_0/libs/statechart/doc/tutorial.html#Stat… but im sure i just miss something any help on this would be great thanks in adavance

2 1

[signals2] Usage (lifespan) question
by Nat Goodspeed 05 Mar '09

05 Mar '09

I'd like suggestions for an API designed around Boost.Signals2. This isn't performance-critical. Assume a publish class containing a boost::signals2::signal. This class's listen() method is a thin wrapper around signal::connect(). Using boost::bind, I can listen() with any method (with appropriate signature) on any class. I'm trying to introduce the concept of filter objects. Each filter is itself a publish object. Client logic composes a chain from some publish object through zero or more filters to the ultimate listener. A filter passes along to its own listeners some subset of the input notifications. The central issue is the lifespan of a filter object. I want the filter objects in such a chain implicitly freed when the ultimate listener disconnects. Approach 1 ========== I could achieve that by managing each filter object with shared_ptr, storing a copy in each listener. When the last listener is destroyed, the filter goes away. (This assumes passing weak_ptr to boost::bind() so the connection itself doesn't keep the listener alive. Further assume I've used connection::track() so that the signal to which the filter is connected implicitly disconnects.) Naturally this effect cascades "upstream" through the filter chain. But it bothers me that such an implementation imposes new requirements on each listener. Beyond declaring a method with the right signature, the listener must now store a shared_ptr to the publish object to which it's listening. That implies a method on the listener to capture the shared_ptr and establish the connection. Therefore, every listener must be derived from some particular base class, and explicit connection and disconnection must be done through the listener rather than directly through the publish object. Either this new inverted API must apply to all publishers/listeners, or client logic must distinguish between the case of connecting to a filter and connecting to any other type of publish object. It feels to me as though this approach has gone into the weeds. Approach 2 ========== Instead of passing a filter's weak_ptr to boost::bind() to the publish object's listen() method, I could instead pass the filter's shared_ptr. Now the publish object upstream from the filter is responsible for the filter's lifespan: when the filter explicitly disconnects, it will (eventually) go away. It will not implicitly disconnect because it won't go away as long as the connection still holds a copy of its shared_ptr. (Frank says[1] he might tweak the library to forget a shared_ptr immediately on disconnect, but currently it survives for a while.) Consider a chain: A is the original publish object B is a filter listening to A C is the ultimate listener, listening to B Again, the goal is to disconnect B (thus deleting it) when C disconnects or is destroyed. We could make B store its connection object to A. The question is, how would B know to disconnect it? How could B become aware of C disconnecting from B's signal object? Approach 3-n ============ Suggestions? Thanks! [1] http://www.nabble.com/Re%3A--signals2--review--The-review-of-the-signals2-l…

1 0

[Serialization] Load an object from the base pointer
by Ryan Fowler 05 Mar '09

05 Mar '09

Hi, How can I read in a from a stream when I don't know the exact type I will be getting? All objects I receive will derive from the same base class, but I won't know when I receive the stream what the derived type will be and therefore I can't create it in advance (like in the docs). class base { ... } class derived1 : public base { ... } class derived2 : public base { ... } int main() { // Get a stream from somewhere boost::archive::binary_iarchive ia( instream ); base b0; derived1 d1; derived2 d2; is >> // derived1 or derived2??? } Ryan Fowler Software Engineer Thomson Reuters Phone: +442075427063 ryan.fowler(a)thomsonreuters.com thomsonreuters.com This email was sent to you by Thomson Reuters, the global news and information company. Any views expressed in this message are those of the individual sender, except where the sender specifically states them to be the views of Thomson Reuters.

2 3

Microsoft compiler bug: Boost 1.38 smart pointers + CLI = crash
by Oleg Golosovskiy 05 Mar '09

05 Mar '09

I have added a new bug to connect.microsoft.com. I can send the sample project if anybody needs it. https://connect.microsoft.com/VisualStudio/feedback/ViewFeedback.aspx?Feedb… Type Bug ID 420897 Status Active Access Restriction Public Opened By OlegGL Blocking Issue No Opened 3/5/2009 Submission Language English ------------------------------------------------------------------------ *Description * The problem looks identical to one described in bug #101889. http://connect.microsoft.com/VisualStudio/feedback/Workaround.aspx?Feedback… There is only one difference - intrusive_ptr instead of shared_ptr originally reported in #101889 The code bellow causes assertion: boost::intrusive_ptr<test_ptr> ptr; if(ptr) assert(false); //MUST NOT HAPPEN BUT IT DOES! Reproduced on VS2005 SP1 and VS2008 SP1 (Windows XP SP2) Boost 1.38 To reproduce the problem, the Boost (booost/smart_ptr.hpp in our case) shall be included in the CLI-compiled file under #pragma managed (push,off) (see the code below) If changed to "#pragma managed (push,on)", then everything is fine Product Language English Version Visual Studio 2008 SP1 Operating System Windows XP Professional Operating System Language English *How to reproduce: *Create any native win32 C++ project, for example a console based. int _tmain(int argc, _TCHAR* argv[]) { do_test(); return 0; } 1) Install and compile BOOST 1.38 2) Add a new CPP file to the project. This file has a minimal intrusive pointer "environment" and a code check. The Code is below. 3) Set /clr option for that particular file (Common Language Runtime Support in the Options) 4) build the project #include "stdafx.h" #include "test3_dll.h" #pragma managed (push,off) #include "boost/smart_ptr.hpp" class intrusive_ptr_base1 { friend void intrusive_ptr_add_ref(intrusive_ptr_base1*); friend void intrusive_ptr_release(intrusive_ptr_base1*); size_t m_ref_counter; void add_ref() { ++m_ref_counter; } void release() { if (--m_ref_counter == 0) delete this; } protected: intrusive_ptr_base1() : m_ref_counter(0) {} virtual ~intrusive_ptr_base1() {} intrusive_ptr_base1(intrusive_ptr_base1 const &) :m_ref_counter(0) {} intrusive_ptr_base1 &operator=(intrusive_ptr_base1 const &) { return *this; } }; inline void intrusive_ptr_add_ref(intrusive_ptr_base1* e) { e->add_ref(); } inline void intrusive_ptr_release(intrusive_ptr_base1* e) { e->release(); } class test_ptr : public intrusive_ptr_base1 { }; #pragma managed (pop) void do_test() { boost::intrusive_ptr<test_ptr> ptr; if(ptr) assert(false); } Actual Results if(ptr) returns true for the empty ptr! Expected Results if(ptr) must return false for the empty ptr TAP Code (if applicable)

2 1

[boost::program_options] Different usage
by Everton Patitucci 05 Mar '09

05 Mar '09

Hello, I'm wondering if I should use boost::program_options and I got some questions about it. All examples that I found are using std::string and std::ostream everytime they're dealing with values being some kind of text... so to display the "usage" of their program they would do something like... cout << desc; 'desc' being a boost::program_options::options_description. or... when adding options to their command line... desc.add_options() ... ("opt", boost::program_options::value<std::string>, "desc") ... What I want to know is, if I didn't want to use std::string nor std::ostream nor std anything in my specific code (I know boost uses it). Is there a not-so-time-expensive way to do it? If there's could someone show me (there's no need for the real code, pseudo is more than OK) if it's supported? something like this for example... desc.add_options() ("opt", bpo::value<char*>, "blabla") and in usage()-like: I don't know if this works, I looked in doc... std::vector< shared_ptr<po::option_description> > opts = desc.options(); std::vector< shared_ptr<po::option_description> >::iterator it = opts.begin(); for (; it != opts.end(); ++it) { shared_ptr<po::option_description> opt = *it; printf("%s\n", opt->format_parameter().c_str()); } Patuti

2 1

[asio] serial port howto?
by Markus Werle 05 Mar '09

05 Mar '09

Hi! For first time users of asio it is not easy to get things started. I found a demo application (minicom_client) on the web, but it uses a raw io_service. In the reference docs I found serial_port_service. What is the difference? Can I use serial_port_service as a drop in replacement where an io_service normally is used? Is there any small example how to use all this? best rgards, Markus

2 1