August 2009 - Boost-users - lists.preview.boost.org

Boost ASIO on Windows Mobile 5, asynchronous functions not working
by James Pritts 28 Aug '09

28 Aug '09

I was able to compile boost asio for Windows Mobile 5 using VC 9.0 and STL Port 5.2.1. The synchronous socket functions seem to be working, but the asynchronous functions never call the provided handlers. Does anyone know if asio's asunchronous socket functions work on WinMo5? I know that WinMo 5 does not have i/o completion ports, but the reactors don't seem to be working either. Is asynchronous i/o not possible on Windows Mobile ?. I used the following functions: client_socket_.async_connect(...); client_socket_.async_read_some(...); The async_connect call establishes a connection, but my connect handler is never called. Thanks, Jimmy.

2 2

[serialization] Linker errors in trunk
by Hartmut Kaiser 28 Aug '09

28 Aug '09

Since the latest changes in trunk I'm seeing linker errors while linking with the serialization library (VS2005 SP1, boost_serialization-vc80-mt-gd-1_40.lib): runtime_support.obj : error LNK2001: unresolved external symbol "__declspec(dllimport) public: __thiscall boost::archive::archive_exception::archive_exception(enum boost::archive::archive_exception::_exception_code,char const *,char const *)" (__imp_??0archive_exception@archive@boost@@QAE@W4_exception_code@012@PBD1@Z) (note the leading underscore in _exception_code) Which is really strange as this symbol is the only missing one. OTOH, the library exports the symbol: boost::archive::archive_exception::archive_exception( enum boost::archive::archive_exception::exception_code, char const *,char const *); (note the missing underscore in exception_code) If I change the first parameter of the archive_exception constructor to be a 'int' everything builds fine. Any clues? Compiler issue? Regards Hartmut

1 0

[lambda] Can I use lambda::bind to a functor with a parametric return type?
by Jesse Perla 28 Aug '09

28 Aug '09

Thanks for the rapid response on the previous question. Now for my real question: I am using an autodifferentiation library and want to use lambda's bind, but still return a generic type. Is this possible? Here is the function I wrote : struct test_function_impl { //result_of supprt template<typename Sig> struct result; template<typename This, typename Arg1, typename Arg2> struct result<This(Arg1, Arg2)> { typedef typename boost::remove_reference<Arg1>::type::value_type type; }; template<typename Vector, typename IntT> typename Vector::value_type //The type in the vector... can be a double or AD type operator()(const Vector& x, IntT y) const{ return x[0] * x[0] + y; } }; The function works on its own, and now std::tr1::result_of seems to work as well. However, I cannot bind to this: using namespace boost::lambda; ublas::bounded_vector<double, 1> x; x[0] = 1.0; int int_value = 1; //THIS FAILS! Not sure if lambda's bind is using result_of? Obviously, can't use the ret<> approach here. auto f = bind(test_function_impl(), _1, int_value); Does lambda support this usage, or should I turn to Phoenix (if so, I am recognizing a trend here). -Jesse

2 1

: Re: How to use the "type" member template pattern for using result_of?
by Jesse Perla 28 Aug '09

28 Aug '09

Thanks to all, Now I was playing around with Phoenix on my binding problem here as well (which is part of the reason for my confusion)... am I right to think that phoenix has a "result" template incompatible with result_of? The following seems to work when using Phoenix lazy functions, etc., but the standard result template did not. struct test_function_impl { //For phoenix... not sure if necessary for template<typename Arg1, typename Arg2> struct result { //typedef typename Arg1::value_type type; }; template<typename Vector, typename IntT> typename Vector::value_type //The type in the vector... can be a double or AD type operator()(const Vector& x, IntT y) const{ return x[0] * x[0] + y; } }; Thanks, Jesse

2 1

Formal Review of Boost.Polygon is halfway through and no actual reviews so far
by Fernando Cacciola 28 Aug '09

28 Aug '09

Dear Developers, The formal review of the Boost.Polygon is halfway through and, while there was some useful discussion, there are still no "actual" reviews (where you clearly vote for inclusion and rejection). If you are interested in a geometric oriented library please try to make the time to post a formal review by the end of September 2 :) Best regards Fernando Cacciola Review Manager

1 0

Linux build of boost using Intel C++ compiler 11.1 or 11.0 and g++ 4.3.3
by Bill O'Hara 28 Aug '09

28 Aug '09

Hi, I guess this may be a known problem but I wonder if there is any straightforward workaround. I need to install boost on our development machines which run Linux and have g++ 4.3.3 installed. We are wondering about also support the Intel c++ compiler (version 11.1). When trying to build boost 1.39 we get many errors during compilation due to g++ builtins like __builtin_isfinite not being found. Example below. Perhaps I've missed something but using the intel toolset when building still seems to involve using these g++ specific features. Is there a workaround? Thanks Bill For example: "icpc" -c -xc++ -O3 -w1 -inline-level=2 -pthread -fPIC -DBOOST_ALL_NO_LIB=1 -DBOOST_MATH_TR1_DYN_LINK=1 -DNDEBUG -I"." -c -o "bin.v2/libs/math/build/intel-linux-11.1/release/threading-multi/lround.o" "libs/math/build/../src/tr1/lround.cpp" ...failed intel-linux.compile.c++ bin.v2/libs/math/build/intel-linux-11.1/release/threading-multi/lround.o... intel-linux.compile.c++ bin.v2/libs/math/build/intel-linux-11.1/release/threading-multi/nextafter.o /usr/include/c++/4.3.3/cmath(522): error: identifier "__builtin_isfinite" is undefined return __builtin_isfinite(__type(__f)); ^ detected during: instantiation of "__gnu_cxx::__enable_if<std::__is_arithmetic<_Tp>::__value, int>::__type std::isfinite(_Tp) [with _Tp=double]" at line 262 of "./boost/math/special_functions/fpclassify.hpp" instantiation of "bool boost::math::detail::isfinite_impl(T, const boost::math::detail::native_tag &) [with T=double]" at line 310 of "./boost/math/special_functions/fpclassify.hpp" instantiation of "bool boost::math::isfinite(T) [with T=double]" at line 56 of "./boost/math/special_functions/next.hpp" instantiation of "T boost::math::float_next(const T &, const Policy &) [with T=double, Policy=c_policies::c_policy]" at line 153 of "./boost/math/special_functions/next.hpp" instantiation of "T boost::math::nextafter(const T &, const T &, const Policy &) [with T=double, Policy=c_policies::c_policy]" at line 27 of "libs/math/build/../src/tr1/c_policy.hpp" instantiation of "T c_policies::nextafter(const T &, const T &) [with T=double]" at line 15 of "libs/math/build/../src/tr1/nextafter.cpp" compilation aborted for libs/math/build/../src/tr1/nextafter.cpp (code 2)

4 4

Re: [Boost-users] [ANN] Boost 1.39.0
by Tom Kent 28 Aug '09

28 Aug '09

I've built windows libraries (vc7.1, vc8, vc9) off of this, if you want them, you can get them from my website: http://boost.teeks99.com Tom > Subject: [Boost-users] [ANN] Boost 1.39.0 > From: Beman Dawes (bdawes_at_[hidden]) > Date: 2009-05-02 16:57:10 > > Boost 1.39.0 is available for download from SourceForge. > > See > http://sourceforge.net/project/showfiles.php?group_id=7586&package_id=8041&… > > New Libraries: Signals2. > > Updated Libraries: Asio, Flyweight, Foreach, Hash, Interprocess, > Intrusive, Program.Options, Proto, PtrContainer, Range, Unordered, > Xpressive. > > Updated Tools: Boostbook, Quickbook. > > Plus the usual bug fixes to numerous libraries. > > Please report any problems to the Boost users or developers mailing > list. > > Thanks, > > --Beman Dawes, Daniel James, and all the other Boosters who help get > releases out

1 1

How to use the "type" member template pattern for using result_of?
by Jesse Perla 28 Aug '09

28 Aug '09

Take the following function: struct test_function { template<typename Vector, typename IntType> typename Vector::value_type operator()(const Vector& x, IntType bound_value) const{ return x[0] * x[0] + bound_value; } }; I wasn't sure exactly how the "type" member template pattern works so that result_of will work. I tried the following, but can't figure out how to extract the type of the arguments or convert to an MPL sequence? Is there some trick I am missing? struct test_function { template<typename Args> struct result { //HOW TO CONVERT TO MPL SEQUENCE OR ANOTHER APPROACH? typedef typename boost::mpl::at_c<Args, 0 >::type::value_type type; }; template<typename Vector, typename IntType> typename Vector::value_type //The type in the vector... can be a double or AD type operator()(const Vector& x, IntType bound_value) const{ return x[0] * x[0] + bound_value; } }; Thanks, Jesse

3 2

[interprocess] remove
by Moritz 28 Aug '09

28 Aug '09

Hi, I have a question concerning the remove function of managed_shared_memory. Actually I am not able to remove a segment of shared memory. See the following example. It writes 0 to the output stream. But why? #include <boost/interprocess/managed_shared_memory.hpp> #include <iostream> int main() { boost::interprocess::managed_shared_memory * managed_shm; managed_shm = new boost::interprocess::managed_shared_memory(boost::interprocess::open_or_create, "TestMem", 1024); delete managed_shm; std::cerr << boost::interprocess::shared_memory_object::remove("TestMem") << std::endl; }

2 3

boost::function unusable as overloadable function argument
by Peter Soetens 28 Aug '09

28 Aug '09

Hi, I had this situation in a bigger program, where a user could supply a string or a function object: <code> #include <boost/function.hpp> #include <string> void foo1(int i, std::string a); void foo1(int i, boost::function<void(void)> foo ); int main(int argc, char** argv) { //boost::function<void(void)> foo = "hello"; // this line won't compile foo1( 3, "hello" ); //error: call of overloaded ‘foo1(int, const char [6])’ is ambiguous return 0; } </code> Although constructing a boost::function from a C string is nonsense, the compiler does consider it as an alternative when trying to find the right foo1. Is there any work around to keep the overloads without requiring the user to write foo1(3, string("hello")); or is this a compiler bug (g++ (Ubuntu 4.3.3-5ubuntu4) 4.3.3)? Thanks for any insights, Peter

3 3