- Boost-users - lists.boost.org

Detect generic base class
by Francesco Biscani 30 Apr '10

30 Apr '10

Hi list, how can I detect if a class inherits from a generic class? E.g., template <class Derived> class base {}; class final: public base<final> {}; I would like to specialise a type trait for all classes deriving from base<whatever>, possibly using enable_if and is_base_of: template <class T, class Enable = void> class trait: boost::false_type {}; // What syntax here? Something like this (which does not work): template <class T, class Derived> class trait<T,typename boost::enable_if<boost::is_base_of<base<Derived>,T> >::type>: boost::true_type {}; The best I've come up with has been putting a typedef for Derived in base, and then using SFINAE: template <class Derived> class base { public: typename Derived derived; }; template <class T> class trait<T,typename boost::enable_if<boost::is_base_of<base<typename T::derived>,T> >::type>: boost::true_type {}; But this is ugly and I'm sure there are much better ways of doing this :) Thanks in advance, Francesco

3 3

error C2661: 'operator new' : no overloaded function takes 3 arguments
by bidhan chandra gupta 30 Apr '10

30 Apr '10

Hi, I have a simple pattern matching console based program something like below: #include void example() { ... boost::regex pattern("[a-zA-Z]"); if (boost::regex_match(line, pattern)){...} ... } While compiling this program in VS2008, I get following error: error C2661: 'operator new' : no overloaded function takes 3 arguments ..\boost_1_35_0\boost\regex\v4\regex_raw_buffer.hpp(177) error C2059: syntax error : '*' ..\boost_1_35_0\boost\regex\v4 \perl_matcher_non_recursive.hpp 99 Need your help to fix this. Thanks

2 1

Parallel(or Distributed) Boost Algorithm
by anil khadka 30 Apr '10

30 Apr '10

I am pretty new to both boost and parallel boost. I am using Brandes Centrality algorithm, but the performance really doesn't meet my expectation. If I used Java, which is only using 1 CPU core would complete it in around 30 hrs. The parallel version which use 8 CPU cores tooks 17 hours!! I don't know why there is just around 13 hours gain if i use 8 cores against 1, unless there are lots and lots of communication between each cores. I am using just 1 machine (not a cluster), right now. Graph size is Node 80K, edges 120K I am reading edges from files. I used distributed compressed_sparse_row to represent graph. Right now, it seems like all the cpu cores are reading files on its own, I am assuming each CPU core will takes its graph part automatically (distributed algorithm do it ?). Note: I downloaded the parallel version from: http://osl.iu.edu/research/pbgl/software/ Any help/suggestions would be really appreciated. Metanil.

1 0

[iostreams] How to make a seekable device seek
by Bo Jensen 30 Apr '10

30 Apr '10

I have a seekable iostream device with the tag set to : typedef boost::iostreams::input_seekable catagory; If I have not defined the functions get and seek, it will not compile, which is expected, so far so good. But when I called it like this : mydevice test("somefile.txt"); boost::iostreams::stream<mydevice> buf(test); buf.seekg(0); it never enters my seek function, which I would expect. I am sure the solution must be obvious. Any pointers here ?

2 3

[units] conversion factor for non-base
by Noah Roberts 30 Apr '10

30 Apr '10

I'm unsure how to go about what I need. I've looked at the library code and the documentation and am not even sure what I want is possible, though it should be. In most of our code we use the si system and have defined composite dimensions for most of what is unique to our application. However, there is one dimension where there is no standard unit in the si system: flow coefficient. The flow coefficient is Q/sqrt(dP), a volumetric flow/sqrt(pressure). We have derived and defined a unit that would be the si standard for this quantity if there was one and things work fine. I have a runtime unit system that then converts between user values that make sense, to our internal si version that does not but works. However, now I want to write some alternative functions that use the actual units that engineers really use: Cv and Kv. I would like these alternative functions to then convert into this si unit we've created but the interface should distinguish between Cv, Kv, and my made up unit. The first part is easy: make a system that only contains this funky dimension and then use it to make quantities that are not in the si system. The second part is what I can't find. The conversion templates in conversion.hpp seem to specifically expect base units which at least the si part is not and my conversions should use my funky si unit as their basis. So the question I have is how can I write conversions such that I can do this: quantity<si::flow_coefficient> fc(42 * Kv); and this: void f(quantity<Kv::coefficient>); void f(quantity<Cv::coefficient>); void f(quantity<si::coefficient>); Where Kv and Cv are base units in their own systems. The problem is sort of like degrees and radians, which I was going to use as a basis when I first was looking at this issue, except that in that case the unit is a base unit for a base dimension and in my case I'm working with a derived dimension within the si system. The only way I can think of to do it is to make a system with the components within the derived dimension and make conversion factors for those. The problem is that I am not sure such systems can be coherent and it would be a significant effort above and beyond the direct conversion factors I already know. -- http://crazyeddiecpp.blogspot.com/

2 4

[Proto] Lazy function test
by Noman Javed 30 Apr '10

30 Apr '10

I am experimenting with the calculator example presented in the proto docs. I am unable to use pow<> when I define it using the make_expr (as preseneted in docs), but when I define it like proto::terminal< struct pow_fun<2> >::type const pow2 = {{}}; it works. I am using it like that in the main calculator_context ctx; ctx.args.push_back(2.0); std::cout << proto::eval( pow2(_1), ctx ) << std::endl; // ok with terminal std::cout << proto::eval( pow2(3.0), ctx ) << std::endl; // ok with terminal How can I make the make_expr version to work? Thanks in anticipation Noman Javed Orléans, France

3 2

[Proto] loosing expression wrappers in transforms
by Robert Valkenburg 30 Apr '10

30 Apr '10

Hi, I have a grammar, domain and an extended expression i.e. struct ga_domain : proto::domain<proto::generator<ga_expr>, ga_grammar> {}; template<typename Expr> struct ga_expr : boost::proto::extends<Expr, ga_expr<Expr>, ga_domain> { // .... }; When the expression is transformed some instances of the ga_expr<> wrapper get removed. For example // some function that expects a ga_expr<Expr> as 2nd arg void evaluate_expr(Res& res, const ga_expr<Expr>& expr) { // .... } // some function which transforms the expression and calls evaluate_expr() template <class Res, class Expr> void evaluate(Res& res, const ga_expr<Expr>& expr) { BOOST_MPL_ASSERT((proto::matches< ga_expr<Expr>, ga_grammar>)); evaluate_expr(res, ET::transform_expression()(expr) ); } With the (do nothing) transform struct transform_expression : proto::nary_expr<proto::_, proto::vararg<transform_expression> > {}; I get the following messages (using gcc 4.4.1, boost 1.42) In function ‘void clifford::evaluate(Res&, const clifford::ga_expr<Expr>&) [with Res = clifford::element<clifford::basis::basis<clifford::cga3::basis::rotator_basis_list, mpl_::integral_c<unsigned int, 105u> >, clifford::cga3::cga3, clifford::default_tags>, Expr = boost::proto::exprns_::expr<clifford::tag::reverse_, boost::proto::argsns_::list1<clifford::ga_expr<boost::proto::exprns_::expr<boost::proto::tag::terminal, boost::proto::argsns_::term<const clifford::element<clifford::basis::basis<boost::mpl::vector10_c<short int, 3, 5, 6, 9, 10, 12, 17, 18, 20, 24>, mpl_::integral_c<unsigned int, 18224744u> >, clifford::cga3::cga3, clifford::default_tags>&>, 0l> > >, 1l>]’: error: no matching function for call to ‘evaluate_expr(clifford::element<clifford::basis::basis<clifford::cga3::basis::rotator_basis_list, mpl_::integral_c<unsigned int, 105u> >, clifford::cga3::cga3, clifford::default_tags>&, boost::proto::exprns_::expr<clifford::tag::reverse_, boost::proto::argsns_::list1<const clifford::ga_expr<boost::proto::exprns_::expr<boost::proto::tag::terminal, boost::proto::argsns_::term<const clifford::element<clifford::basis::basis<boost::mpl::vector10_c<short int, 3, 5, 6, 9, 10, 12, 17, 18, 20, 24>, mpl_::integral_c<unsigned int, 18224744u> >, clifford::cga3::cga3, clifford::default_tags>&>, 0l> >&>, 1l>&)’ So the clifford::ga_expr<> wrapper got removed from the unary reverse operator and i'm not sure what i'm supposed to add to make it stay. thanks

3 5

Why is basic_stream_socket::get_io_service not const?
by Christopher Yeleighton 30 Apr '10

30 Apr '10

According to the documentation [1], the member function get_io_service has the following synopsis: boost::asio::io_service & get_io_service(); Why is this function not constant? What does it change within the socket? [1] <URL:http://www.boost.org/doc/libs/1_41_0/doc/html/boost_asio/reference/basi…>

1 0

Re: [Boost-users] [BGL] Out-of-range Edge Endpoints
by Hossein Haeri 30 Apr '10

30 Apr '10

Dear Andrew, > > > What I can't see is what will happen when the endpoints don't fall in > > > [0, n). Is that documented somewhere? > > > > > > > I don't think so. It's basically undefined behavior -- it usually ends > > up in an out-of-bounds array access and thus likely a crash. > > In the case of the constructor, it may. And, my question was exactly about the ctor. I'm surprised to see there is no exact behviour specification for this. Shouldn't it be done sometime? > In the case of add_edge (with the vertex list == vecS), adjacency_list > will automatically resize -- I think I remember being surprised with > this before :) The documentation says: "If the VertexList selector is vecS, and if either vertex descriptor u or v (which are integers) has a value greater than the current number of vertices in the graph, the graph is enlarged so that the number of vertices is std::max(u,v) + 1." Cheers, --Hossein

1 0

[BGL] Out-of-range Edge Endpoints
by Hossein Haeri 29 Apr '10

29 Apr '10

Dear all, In the following documentation page for adjacency_list: http://www.boost.org/doc/libs/1_42_0/libs/graph/doc/adjacency_list.html according to what I read below the following ctor, endpoints of edges must all fall in [0, n): template <class EdgeIterator> adjacency_list(EdgeIterator first, EdgeIterator last, vertices_size_type n, edges_size_type m = 0, const GraphProperty& p = GraphProperty()) What I can't see is what will happen when the endpoints don't fall in [0, n). Is that documented somewhere? TIA, --Hossein

3 2