November 2009 - Boost-users

[asio] why no random_access_handle on non-windows platforms?
by Zachary Turner 04 Nov '09

04 Nov '09

I notice there's a random_access_handle on windows, and is needed in order to use functions like async_read_some_at. Am I missing some obvious reason why there is not an analagous random access handle for posix systems? I guess I can define my own posix::random_access_handle, posix::basic_random_access_handle, posix::random_access_handle_service just like is done for windows, but i don't see why there is anything windows-specific about random access handles in the first place. Ultimately I need to be able to perform random access asynchronous file i/o on both linux and windows using a single codebase. Any advice appreciated.

1 0

boost::interprocess communication between windows service and user application
by jmpulido 04 Nov '09

04 Nov '09

Hi, I'm pretty new to boost::interprocess but I've been performing some successful communication tests between two applications, using boost::interprocess::shared_memory_object and boost::interprocess::mapped_region. But when one of those processes is a windows service, then I get a file not found exception (I've catched it and dumped it into a file), just after initializing a shared_memory_object. It seems that the shared memory is actually managed by means of a file in disk that probably the user is creating in a location and the service in a different one. I've been googling this issue for hours, but the only resource I was able to find is this conversation: http://stackoverflow.com/questions/488877/boostinterprocess-between-windows… which has not been very helpful since I have no clue on how to use boost::interprocess::managed_windows_shared_memory or even if it could serve to my purpose somehow. Thank you.

2 1

[mpl, preprocessor] Using BOOST_MPL_ASSERT in BOOST_PP_REPEAT
by Jean-Louis Leroy 03 Nov '09

03 Nov '09

Consider: #define BOOST_PP_RDB_MPL_ASSERT(z, n, t) \ BOOST_MPL_ASSERT((BOOST_PP_TUPLE_ELEM( \ 2, 0, t)<BOOST_PP_CAT(BOOST_PP_TUPLE_ELEM(2, 1, t), n)>)); // in a file passed to BOOST_PP_ITERATE() BOOST_PP_REPEAT(n, BOOST_PP_RDB_MPL_ASSERT, (is_column_container, Col)) This works with msvc9 express but not with g++ 4.4.1. When I dig it I see that BOOST_MPL_ASSERT creates an identifier based on BOOST_MPL_AUX_PP_COUNTER() which in turns uses either __COUNTER__ (with msvc9) or __LINE__ (with g++). So with g++ we get name clashes when `n` is greater than one, because the ASSERT macro attempts to create several variables with the same name in the same scope. I guess I could copy the code of BOOST_MPL_ASSERT() in my BOOST_PP_RDB_MPL_ASSERT macro and incorporate `n` in the identifier. Is there a better solution ? Thanks, J-L

1 0

ENH RQST: chdir in boost::filesystem
by Andrew Marlow 03 Nov '09

03 Nov '09

I have recently moved back to C++ from a 2 year spell with Java. The C++ code I am working on is for Windoze with Visual Studio 8. We are using boost 1.34.1. The prog I am working on has to do a few filesystem things so I have started to use boost::filesystem. However, there does not seem to be a way to change the current working directory using boost::filesystem. So I am forced to use chdir, which is _chdir on Windoze. On Unix the header for the prototype is in unistd.h, on Windoze it is in direct.h. So its back to good old ifdefs. Suddenly, java looks more attractive. Ifdefs are just so nostalgic.... Is there any chance this can be added to boost please? I realise that there may be more to it than just wrapping the chdir/_chdir call. There is the whole issue of return values versus exceptions, errno etc etc. I do not know what the current thinking is about these things in boost. Perhaps someone can enlighten me? -- Regards, Andrew M. http://www.andrewpetermarlow.co.uk

3 2

[date_time] Creating a ptime from a double
by Ryan McConnehey 03 Nov '09

03 Nov '09

A lot of the constructors of ptime deals with getting the current system time or time structures. I have a double with the seconds and fractional microseconds from midnight January 1, 1970. What is the best way to create a ptime object from this double? Ryan

2 4

Template operator() overloading for types in a mpl::vector
by Olivier Tournaire 03 Nov '09

03 Nov '09

Hi all, I am currently trying to write a simple functor, and I would like to have the same overload for the operator () for all types in a mpl::vector. // Types definitions // Gray image types typedef boost::mpl::vector< boost::gil::gray8_image_t , boost::gil::gray16_image_t , boost::gil::gray32_image_t, boost::gil::gray32F_image_t , boost::gil::gray64F_image_t > gray_image_types; // RGB image types typedef boost::mpl::vector< boost::gil::rgb8_image_t , boost::gil::rgb16_image_t , boost::gil::rgb32_image_t > rgb_image_types; // Functor to compute min / max of an image over channels struct pixel_compare_less { template <typename PixelType> bool operator()( const PixelType& p1 , const PixelType& p2 ) const { return at_c<0>(p1) < at_c<0>(p2); } }; struct any_view_min_max { typedef std::pair<float, float> result_type; template <typename View> result_type operator()(const View& v) const { typedef typename View::iterator iterator; std::pair< iterator, iterator > result = boost::minmax_element( v.begin() , v.end() , pixel_compare_less() ); return std::make_pair( *(result.first) , *(result.second) ); } }; This works fine for 'gray_image_types', but need to be specialized for types in rgb_image_types. The specialization is the same, so, I am looking for an automated way to do this. "Manually", I can write this: template <> any_view_min_max_optimized::result_type any_view_min_max_optimized::operator()<rgba8_view_t>(const rgb8_view_t& v) const { // ... } template <> any_view_min_max_optimized::result_type any_view_min_max_optimized::operator()<rgba16_view_t>(const rgb16_view_t& v) const { // ... } And so on ... Could you help me to have an efficient way to do this for all types 'rgb_image_types' without writing each specialization ? Regards, Olivier

2 4

[RegEx] Linking issues with boost 1.40.0 included directly in our tree
by Conrad Whelan 03 Nov '09

03 Nov '09

Hi, we use boost in our build, however, instead of using all of the Boost.Build and linking infrastructure, we integrate the source files directly into our build. We've always done things this way and had no problem. The implicit linking is a bit of a pain. However, now I have a strange issue. I am currently upgrading from 1.39.0 to 1.40.0, and am getting linking issues in the regex library. In particular, the linker is getting undefined references for boost::re_detail::perl_matcher in files where we create boost::regex and boost::regex_match objects. When I look through the source for the perl_matcher template, I can find it in the boost/regex/v4/perl_matcher.hpp include file, which we include within our build. I've also included the src/ and performance/ directories from boost/libs/regex in our build, and the linking still doesn't work. Perhaps someone has an incite into any preprocessor flags that I need to set in order to create an implementation for these missing functions in 1.40.0? Note, we had this working fine for 1.39.0, and I need to upgrade for a bugfix for a different boost library. thanks in advance, conrad

2 1

Re: [Boost-users] Template operator() overloading for types in a
by Tan, Tom (Shanghai) 03 Nov '09

03 Nov '09

>And so on ... Could you help me to have an efficient way to do this for all >types 'rgb_image_types' without writing each specialization ? Your situation seems similar to the one discussed here: http://stackoverflow.com/questions/1492204/is-it-possible-to-generate-ty pes-with-all-combinations-of-template-arguments. I actually refined the idea and modeled the mpl::for_each, and came up with this code pasted below for your reference.What you need to modify is to reduce the template arguments from 3 to 2 to suit your needs, hope it helps. #ifndef FOR_EACH_CARTESIAN_PRODUCT_H #define FOR_EACH_CARTESIAN_PRODUCT_H #include <boost/mpl/vector.hpp> #include <boost/mpl/begin.hpp> #include <boost/mpl/end.hpp> #include <boost/mpl/deref.hpp> #include <boost/type_traits/is_integral.hpp> #include <iostream> #include <typeinfo> /* usage 1: typedef recursive::for_each_cartesian-product< stock::map_containers, stock::shrinker_equal_filters, stock::shrinker_method_tags, RunnableBody > benchmark_type; benchmark_type::run_all(); usage 2: typedef recursive::for_each_cartesian-product< stock::map_containers, stock::shrinker_equal_filters, stock::shrinker_method_tags, RunnableBody >::each_product<> benchmark_type; benchmark_type::run(); */ namespace recursive { using boost::is_same; using boost::is_integral; using boost::mpl::begin; using boost::mpl::end; using boost::mpl::next; using boost::mpl::if_; using boost::mpl::deref; using boost::mpl::size; using boost::mpl::advance; using boost::mpl::distance; using boost::mpl::int_; using boost::mpl::eval_if_c; namespace detail { static unsigned int total_recursions = 0; } // generate a Cartesian Product of 3 type sequences and apply RunnableBody on each generated type template < class UTypes, // Forward Sequence, e.g. boost::mpl::vector class VTypes, // Forward Sequence, e.g. boost::mpl::vector class WTypes, // Forward Sequence, e.g. boost::mpl::vector class RunnableBody // class type that has a nested templated run() member function > struct for_each_cartesian_product { // forward declaration template < class UIterator, class VIterator, class WIterator > class each_product; struct end_of_recursion_tag { static void run(RunnableBody& f) { #ifndef NDEBUG std::cout << "End of " << detail::total_recursions << " recursions: RunnableBody = " << typeid(RunnableBody).name() << std::endl; #endif detail::total_recursions = 0; } }; // convenient interface to run RunnableBody on all generated types static void run_all(RunnableBody& f = RunnableBody()) { each_product<>::run(f); } // this class implements recursion body template < class UIterator, class VIterator, class WIterator > struct next_product { // u_begin is not necessary ;) // it would be cheaper not to pre-declare all of them since we force evaluation // however this dramatically increase the readability typedef typename begin<VTypes>::type v_begin; typedef typename begin<WTypes>::type w_begin; typedef typename end<UTypes>::type u_end; typedef typename end<VTypes>::type v_end; typedef typename end<WTypes>::type w_end; typedef typename next<UIterator>::type u_next; typedef typename next<VIterator>::type v_next; typedef typename next<WIterator>::type w_next; typedef typename if_< is_same<typename w_next, w_end>, typename if_< is_same<v_next, v_end>, typename if_< is_same<u_next, u_end>, end_of_recursion_tag, each_product< u_next, v_begin, w_begin > >::type, each_product< UIterator, v_next, w_begin > >::type, each_product< UIterator, VIterator, w_next > >::type type; }; // interface // this class run test on generated types in thos round and go to next*/ template < class UIterator = typename begin<UTypes>::type, class VIterator = typename begin<VTypes>::type, class WIterator = typename begin<WTypes>::type > struct each_product { // publc accessible internals of this type typedef typename each_product< UIterator, VIterator, WIterator > this_type; // types typedef typename deref<UIterator>::type UType; typedef typename deref<VIterator>::type VType; typedef typename deref<WIterator>::type WType; // index types typedef typename distance<typename begin<UTypes>::type, UIterator>::type UIndexType; typedef typename distance<typename begin<VTypes>::type, VIterator>::type VIndexType; typedef typename distance<typename begin<WTypes>::type, WIterator>::type WIndexType; // core static void run(RunnableBody& f = RunnableBody()) { // increment recursion counter ++detail::total_recursions; //f.template <this_type>(); f.operator()<this_type>(); // generate <<next>> target type and go to the next round of recursion typedef typename next_product< UIterator, VIterator, WIterator >::type next_type; next_type::run(f); } }; // alternative interface template < size_t UIndex = 0, size_t VIndex = 0, size_t WIndex = 0 > struct each_product_by_index { static_assert(UIndex <= size<UTypes>::type::value, "too large stock container index"); static_assert(VIndex <= size<VTypes>::type::value, "too large stock shrinker_group_number index"); static_assert(WIndex <= size<WTypes>::type::value, "too large stock shrinker_group_size index"); typedef typename begin<UTypes>::type u_begin; typedef typename begin<VTypes>::type v_begin; typedef typename begin<WTypes>::type w_begin; static void run(RunnableBody& f = RunnableBody()) { // delegate to iterator-version each_product< typename advance<u_begin, int_<UIndex> >::type, typename advance<v_begin, int_<VIndex> >::type, typename advance<w_begin, int_<WIndex> >::type >::run(f); } }; }; // an example test function class impl. struct print_typeid { template< class T> void operator()() const { typedef typename T::UType U; typedef typename T::VType V; typedef typename T::WType W; typedef typename T::UIndexType UIndex; typedef typename T::VIndexType VIndex; typedef typename T::WIndexType WIndex; // print the typeinfo std::cout << "this is a demo implementation.\n"; std::cout << detail::total_recursions << "[" << UIndex::value ", " << VIndex::value ", " << WIndex::value << "]" << ": " << typeid(U).name() << "," << typeid(V).name() << "," << typeid(W).name() << std::endl; } }; }// namespace recursive_test

1 0

Re: [Boost-users] Boost::Pool - Singleton_pool
by Paul Byrne 03 Nov '09

03 Nov '09

> From: Ondrej Sluciak <ondrej.sluciak(a)nt.tuwien.ac.at> > To: boost-users(a)lists.boost.org > Subject: [Boost-users] Boost::Pool - Singleton_pool > Message-ID: <4AEEFEED.7030905(a)nt.tuwien.ac.at> > Content-Type: text/plain; charset=ISO-8859-1; format=flowed > > Hi, > Is there a way on how to derive a class from pool::singleton_pool class? > Something like: > > template<typename T> > genericsingleton : boost::singleton_pool<boost::pool_allocator_tag, > sizeof(T)> > > If I want to do this I always get this error: > > Error 1 error C2248: > 'boost::singleton_pool<Tag,RequestedSize>::singleton_pool' : cannot > access private member declared in class > 'boost::singleton_pool<Tag,RequestedSize>' > c:\mycode\boosttest\boosttest\genericsingleton.h 23 > > Thank you for answer. > Hi Ondrej, I guess that you are getting this error when you try to construct an instance of your genericsingleton<> class. I don't believe that you will be able to do this because your base class has a private constructor. I suppose if you explain what your purpose is (by deriving from singleton_pool) then someone might be able to suggest a solution for you. Regards, Paul

1 0

Interrupting Socket connect()
by Simon Reye 02 Nov '09

02 Nov '09

If I have a socket like so: boost::asio::ip::tcp::socket tcpSocket; and call connect in a boost::thread like so: tcpSocket.connect(endPoint, error); It will 'hang' for up to 30 seconds if it can't immediately connect to the end point. If I interrupt() and join() the boost thread it has to wait for the connect() call to finish. Is there a way to cancel or interrupt the connect() call? Thanks Simon

2 1