- Boost-users - lists.boost.org

asio: how to handle disk IO, recommendations
by Avi Bahra 23 Oct '12

23 Oct '12

Currently I has a single threaded server based on boost ASIO. This works correctly most of the time. However on systems where i know that the disk's are slow. , the server grinds to a halt, when dealing with the blocking disk IO. I could not find any examples or documentation in ASIO, in how to get round these issues. It appears that ASIO does not provide any support for async' disk IO. So I assume this is handled by use of threads ? Are there any examples of using ASIO with threaded disk IO, since this must be very common use case ? Is the strand functionality appropriate for handing disk IO, in an async' manner. Any recommendations/advice on this issues would be greatly appreciated. Best regards, Ta, Avi

2 1

Boost install creates no boost library files
by niallm 22 Oct '12

22 Oct '12

Hi guys, I'm running OS X 10.8.2 and I'm trying to install boost. I'm using Enthought python. So I've unpacked the tar file boost_1_51_0.tar.bz2 into usr/local. Then in the resulting directory boost_1_51_0, I do the following commands: $ ./bootstrap.sh --with-python=2.7 --with-python-root=../../Library/Frameworks/EPD64.framework /Versions/Current/bin/python then: $ ./b2 link=shared --with-python --with-math (I only need the python and math parts). This returns: he Boost C++ Libraries were successfully built! The following directory should be added to compiler include paths: /usr/local/boost_1_51_0 The following directory should be added to linker library paths: /usr/local/boost_1_51_0/stage/lib Then I did: $ sudo ./b2 --prefix=usr/local link=shared --with-python --with-math install This returns: common.mkdir usr common.mkdir usr/local common.mkdir usr/local/include common.mkdir usr/local/include/boost Followed by lots of lines starting 'common.copy...' and finished with '...updated 9752 targets...'. However when I check in usr/local/include there is no boost directory, and there are no boost libraries in usr/local/lib Thanks in advance -- View this message in context: http://boost.2283326.n4.nabble.com/Boost-install-creates-no-boost-library-f… Sent from the Boost - Users mailing list archive at Nabble.com.

3 2

[graph] Extending bgl_named_params
by Philipp Moeller 22 Oct '12

22 Oct '12

Hi, I maintain a library that introduces its own graph properties for domain specific algorithms and provides it's own implementation of bgl_named_parameters by essentially mimicking bgl_named_parameters one-to-one. With the refactoring in 1.51 that implementation broke and I had to add another implementation of named parameters which essentially looks like this: enum my_property_t { my_property }; template <typename T, typename Tag, typename Base = boost::no_property> struct foo_bgl_named_params : boost::bgl_named_params<T, Tag, Base> { typedef boost::bgl_named_params<T, Tag, Base> base; typedef foo_bgl_named_params self; foo_bgl_named_params(T v = T()) : base(v) {} foo_bgl_named_params(T v, const Base& b) : base(v, b) {} template <typename PMap> foo_bgl_named_params<PMap, my_property_t, self> my_property_map(const PMap& p) const { typedef foo_bgl_named_params<PMap, my_property_t, self> Params; return Params(p, *this); } }; // the respective free functions... template <typename PMap> foo_bgl_named_params<PMap, my_property_t, self> my_property_map(const Pmap& p) const { typedef foo_bgl_named_params<PMap, my_property_t, self> Params; return Params(p); } // partial specialization for lookup_named_param_def namespace boost { template <typename T, typename Tag, typename Base, typename Def> struct lookup_named_param_def<Tag, foo_bgl_named_params<T, Tag, Base>, Def> { typedef T type; static const type& get(const bgl_named_params<T, Tag, Base>& p, const Def&) { return p.m_value; } }; template <typename Tag1, typename T, typename Tag, typename Base, typename Def> struct lookup_named_param_def<Tag1, foo_bgl_named_params<T, Tag, Base>, Def> { typedef typename lookup_named_param_def<Tag1, Base, Def>::type type; static const type& get(const bgl_named_params<T, Tag, Base>& p, const Def& def) { return lookup_named_param_def<Tag1, Base, Def>::get(p.m_base, def); } }; } // boost This is brittle because it the type returned by the inherited functions is not the same and chaining will not work if one of those functions is called first, e.g. all custom properties have to be chained together first. >From the documentation it is not really obvious to me if other libraries are even supposed to provide their own properties and how/if they actually should be able to extend the named parameters API. Is that simply not an intended feature of the BGL or can that be done differently? Cheers, Philipp

2 1

Linking issue with Boost 1.44 and GCC 4.6
by Joseph Sulewski 21 Oct '12

21 Oct '12

Hi, I'm having a very strange compilation issue with Boost. I've used boost in other C++ projects in the past and have had no issues. I'm writing a program to compile against Hypertable which relies on the boost 1.44 version of the library. I'm writing this on Intel Linux and compiling with GCC 4.6.3 Here is my linker command. I removed the .o's from my application to make it simpler to read: g++ -L/opt/hypertable/current/lib -L/home/l3/development/libraries/boost_1_44_0/stage/lib -pthread -o "AppLogger" MYDOT-O-FILES.o -lboost_system -lboost_iostreams -lboost_filesystem -lboost_program_options -lboost_thread -lsigar-x86-linux -lz -lHypertable -lHyperspace -lHyperTools -lHyperComm -lHyperCommon As you can see I'm including a number of boost libraries but I'm getting a lot of very strange errors. For example, opt/hypertable/current/lib/libHypertable.a(Client.cc.o): In function `Hypertable::Client::Client(std::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, unsigned int)': Client.cc:(.text+0x26a5): undefined reference to `boost::program_options::variables_map::variables_map()' >From what I can see the precompiled library from Hypertable is not finding the functions in the boost library. Here is another I get Client.cc:(.text+0x26a5): undefined reference to `boost::program_options::variables_map::variables_map()' /opt/hypertable/current/lib/libHypertable.a(Client.cc.o): In function `Hypertable::Client::Client(std::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, unsigned int)': I get a bunch of these: HqlInterpreter.cc:(.text._ZN5boost9call_onceIPFvvEEEvRNS_9once_flagET_[void boost::call_once<void (*)()>(boost::once_flag&, void (*)())]+0x151): undefined reference to `boost::detail::once_epoch_cv' There are many many more linker errors but I didn't want to clog up the system with all the errors. Any ideas would be greatly appreciated Thanks

3 5

[proto] transform with calling templates
by Felipe Magno de Almeida 21 Oct '12

21 Oct '12

Hello, I had the following transform and function object: struct bind_functions_transform : proto::when < proto::terminal<binding::placeholder::method_value<proto::_,proto::_> > , bind_method_call(proto::_value, proto::_data, proto::_state)> > struct bind_method_call : proto::callable { template <typename T> struct result { typedef typename boost::function_types::parameter_types<T>::type parameter_types; typedef typename boost::mpl::next <typename boost::remove_reference <typename boost::mpl::at_c<parameter_types, 2u>::type>::type>::type type; }; template <typename Sig, typename F, std::size_t I, std::size_t N> boost::mpl::size_t<I+1> operator()(binding::placeholder::method_value<Sig, F>const& method , boost::fusion::vector<Class&, environment, binding::virtual_table<N>&>& data , boost::mpl::size_t<I> v) const { ... } }; So, I needed one more compile-time information which would be equal to all calls and transforms, so I made my transform into a template and also bind_method_call as follows: template <typename T> struct bind_functions_transform : proto::when < proto::terminal<binding::placeholder::method_value<proto::_,proto::_> > , bind_method_call<T>(proto::_value, proto::_data, proto::_state)> > template <typename T> struct bind_method_call : proto::callable [...] And then I received the following error: ../boost_1_50_0/boost/proto/transform/detail/preprocessed/construct_funop.hpp:24:33: error: no matching function for call to ‘jvb::binding::bind_method_call<hello_world>::bind_method_call(const jvb::binding::placeholder::method_value<void(), void (hello_world::*)()>&, boost::fusion::vector<jvb::class_&, jvb::environment, jvb::binding::virtual_table<1ul>&, boost::fusion::void_, boost::fusion::void_, boost::fusion::void_, boost::fusion::void_, boost::fusion::void_, boost::fusion::void_, boost::fusion::void_>&, const mpl_::size_t<0ul>&)’ I've remembered the workaround for msvc with proto::call. Because it seemed that the transform was actually trying to construct instead of calling bind_method_call, so I changed to: template <typename PeerClass> struct bind_functions_transform : proto::when < proto::terminal<binding::placeholder::method_value<proto::_,proto::_> > , proto::call<bind_method_call<PeerClass>(proto::_value, proto::_data, proto::_state)> > And it worked. Well, my question is: is this a known limitation? I didn't find it in proto documentation. I'm using GCC 4.7.1 and x86_64 with Linux Regards, -- Felipe Magno de Almeida

2 1

Boost python, unique_ptr and abstract class
by Maxim Yegorushkin 19 Oct '12

19 Oct '12

I am using boost::python to expose a C++ API into Python. Here is what I am trying to do: #include <boost/python.hpp> #include <memory> struct Subscriber { virtual ~Subscriber() = 0; static std::unique_ptr<Subscriber> create(); }; template<class T> inline T* get_pointer(std::unique_ptr<T> const& p) { return p.get(); } BOOST_PYTHON_MODULE(subscriber) { // Expose the class. boost::python::class_< Subscriber , std::unique_ptr<Subscriber> , boost::noncopyable >("Subscriber"); // Expose its factory function. bp::def("createSubscriber", Subscriber::create); } boost::python::class_<> constructor invocation spits out errors (full compiler log attached): /usr/local/ots/boost-1.51.0/include/boost/python/converter/as_to_python_function.hpp:40:75: error: use of deleted function ‘std::unique_ptr<_Tp, _Dp>::unique_ptr(const std::unique_ptr<_Tp, _Dp>&) [with _Tp = Subscriber; _Dp = std::default_delete<Subscriber>; std::unique_ptr<_Tp, _Dp> = std::unique_ptr<Subscriber>]’ /usr/local/ots/boost-1.51.0/include/boost/python/object/pointer_holder.hpp:194:14: error: cannot allocate an object of abstract type ‘Subscriber’ How do I make boost::python friendly to std::unique_ptr<> and how do I wrap an abstract class please? -- Maxim

1 0

Occasional blocking with tcp::socket::receive
by Stuart Siegel 19 Oct '12

19 Oct '12

The call to receive below will occasionally hang. All interactions are synchronous and single threaded. Things work perfectly if I use the alternative <sys/socket.h> (POSIX) implementation under the #else. Since I'm pretty new to boost::asio, I must be doing something wrong but after many hours of trying different things, I'm still stuck. Any advice would be appreciated. #if 1 namespace bas = boost::asio; remaining = X //(something typically about 5000 bytes of char data) bas::streambuf b; bas::streambuf::mutable_buffers_type bufs = b.prepare(1024); while (remaining > 0) { size_t n = sock.receive(bufs); //<-------------this blocks every so often remaining -= n; for ( bas::buffers_iterator<bas::streambuf::mutable_buffers_type, char> it = bas::buffers_begin(bufs); it < (bas::buffers_begin(bufs) + n); it++ ) qba.append(*it); } #else //this works perfectly in the 250000 or so test calls I've hit it with char buf[1024]; while (remaining > 0) { n = read( sock, buf, min( sizeof(buf), remaining ) ); remaining -= n; for (int i = 0 ; i < n; i++ ) qba.append(buf); } #endif

1 0

[Boost.Functional.OverloadedFunction][C++03] Number of overloads in boost::overload_function<...>
by Felipe Magno de Almeida 19 Oct '12

19 Oct '12

Hello, Is there a way to know how many overloads are there in a overloaded_function without resorting to preprocessor metaprogramming in C++03? It doesn't seem to be a MPL sequence, for example. Regards, -- Felipe Magno de Almeida

3 6

boost solution for transforming stl algorithms on proxy iterators?
by Oswin Krause 19 Oct '12

19 Oct '12

Hi, I often have problems which look a bit like this: //example code, only for problem statement //std::vector<bool> chosen only because it is no container :). If this is too silly for you, //think about a matrix interpreted as range of row-vectors. std::vector<bool> myContainer;//does only have proxy iterators. std::vector<double> myDistances; //stores for every element in myContainer a distance //now i want to to sort the elements of both ranges in place //as if they were in a hypothetic range std::vector<std::pair<bool, double> > //and sort them by the second part: boost::sort(make_key_value_range(myContainer,myDistances));//bam, not allowed I omited the definition of make_key_value_range but it should be clear that this returns a proxy range with writable iterators and reference type similar to std::pair<std::vector<bool>::reference,double&> for which swap is overloaded to do the right thing. This is not allowed by the standard as boost::sort calls std::sort which in turn requires random_access_iterators - and they disallow proxy references. This issue is AFAIK resolved with C++11 but some of us still need to support C++03 - and I am really annoyed by this problem. The thing is: this will compile on most compilers. One could just lie about the iterator category, saying it is a random_access_iterator and even VC would be happy with it. However it could just break randomly on a new compiler without warning or has severe performance issus (for example gcc works, but completely ignores my nicely defined swap and instead does a triangle copy :-( ). So how can one resolve this issue? I have the problem, that i need to do this in-place, since copying would result in std::bad_alloc almost surely. Also the above single line of code just looks that nice, that i just want to have it this way. So, how do you resolve this issue? is there maybe some boost library capable of getting this done? e.g. boost.betterSTL? :-). Greetings, Oswin

5 10

Using Adaptable Generator concept with generate_n and boost smart pointers
by Marco Craveiro 18 Oct '12

18 Oct '12

Hi boosters, I'm trying to create a class that models the concept of adaptable generator [1]. The idea is that it generates pointers to test data objects which I'd like to then put inside of smart pointers such as shared_ptr or scoped_ptr (or even std::shared_ptr) via generate_n, etc. This is a (non-compilable) example of my intentions: #include <algorithm> #include <iostream> #include <memory> #include <vector> #include <iterator> struct base { int i; }; struct derived : public base { int z; }; struct generator { typedef base result_type; result_type* operator()() { return new derived(); } }; int main(void) { std::vector<std::shared_ptr<base> > z; std::generate_n(std::back_inserter(z), 5, generator()); return 0; } The std::shared_ptr is incidental; it could have been boost::shared_ptr, etc - its up to client code to decide. I get this to work fine when result_type is say boost::shared_ptr<base> but what I really like is for it to be a raw pointer; is there some boost magic (or standard C++ 11 magic) I can apply to convert it to the value_type of the container I'm trying to generate? Many thanks for your time, Marco -- So young, and already so unknown -- Pauli blog: http://mcraveiro.blogspot.com [1] http://www.sgi.com/tech/stl/AdaptableGenerator.html

1 0