- Boost-users - lists.boost.org

Invitation to connect on LinkedIn
by Alexander Dong Back Kim 10 Aug '10

10 Aug '10

LinkedIn ------------Alexander Dong Back Kim requested to add you as a connection on LinkedIn: ------------------------------------------ Hermann, I'd like to add you to my professional network on LinkedIn. - Alexander Dong Back Accept invitation from Alexander Dong Back Kim http://www.linkedin.com/e/9m2kgu-gcnzwgc9-6e/vyhGWmEqB2IKeRN9DMTel5y44rvJcs… View invitation from Alexander Dong Back Kim http://www.linkedin.com/e/9m2kgu-gcnzwgc9-6e/vyhGWmEqB2IKeRN9DMTel5y44rvJcs… ------ (c) 2010, LinkedIn Corporation

1 0

lexical_cast
by Internet Retard 09 Aug '10

09 Aug '10

Will lexical_cast be included in the next C++ standard? Is there a list someplace that covers everything from boost that is supposed to make it into the next standard? Thanks

2 1

compiling chapter 2 BGL
by Louis Lavery 09 Aug '10

09 Aug '10

I've owned the BGL book for a few years and am, just now, having a go at chapter 3. I've written... #include <boost/graph/adjacency_list.hpp> int main() { return EXIT_SUCCESS; } ...and, when I compile it using vc9, get... Compiling... main.cpp c:\boost_1_42_0\boost\unordered\unordered_set.hpp(402) : error C2220: warning treated as error - no 'object' file generated c:\boost_1_42_0\boost\unordered\unordered_set.hpp(500) : see reference to class template instantiation 'boost::unordered_set<T,H,P,A>' being compiled c:\boost_1_42_0\boost\unordered\unordered_set.hpp(402) : warning C4224: nonstandard extension used : formal parameter 'hash' was previously defined as a type c:\boost_1_42_0\boost\unordered\unordered_set.hpp(890) : warning C4224: nonstandard extension used : formal parameter 'hash' was previously defined as a type c:\boost_1_42_0\boost\unordered\unordered_set.hpp(989) : see reference to class template instantiation 'boost::unordered_multiset<T,H,P,A>' being compiled c:\boost_1_42_0\boost\multi_index\detail\bucket_array.hpp(115) : warning C4224: nonstandard extension used : formal parameter 'hash' was previously defined as a type c:\boost_1_42_0\boost\multi_index\detail\bucket_array.hpp(165) : see reference to class template instantiation 'boost::multi_index::detail::bucket_array<Allocator>' being compiled c:\boost_1_42_0\boost\multi_index\hashed_index.hpp(437) : warning C4224: nonstandard extension used : formal parameter 'hash' was previously defined as a type c:\boost_1_42_0\boost\multi_index\hashed_index.hpp(1170) : see reference to class template instantiation 'boost::multi_index::detail::hashed_index<KeyFromValue,Hash,Pred,SuperMeta,TagList,Category>' being compiled c:\boost_1_42_0\boost\multi_index\hashed_index.hpp(462) : warning C4224: nonstandard extension used : formal parameter 'hash' was previously defined as a type c:\boost_1_42_0\boost\multi_index\hashed_index.hpp(492) : warning C4224: nonstandard extension used : formal parameter 'hash' was previously defined as a type c:\boost_1_42_0\boost\graph\named_graph.hpp(304) : warning C4224: nonstandard extension used : formal parameter 'property' was previously defined as a type c:\boost_1_42_0\boost\graph\named_graph.hpp(316) : see reference to class template instantiation 'boost::graph::named_graph<Graph,Vertex,VertexProperty>' being compiled c:\boost_1_42_0\boost\graph\named_graph.hpp(309) : warning C4224: nonstandard extension used : formal parameter 'property' was previously defined as a type c:\boost_1_42_0\boost\graph\named_graph.hpp(360) : warning C4224: nonstandard extension used : formal parameter 'property' was previously defined as a type c:\boost_1_42_0\boost\graph\named_graph.hpp(368) : warning C4224: nonstandard extension used : formal parameter 'property' was previously defined as a type Build log was saved at "file://c:\sigmaTrunk\boost\vc\Debug\BuildLog.htm" boost - 1 error(s), 10 warning(s) ========== Build: 0 succeeded, 1 failed, 0 up-to-date, 0 skipped ========== ...is this to be expected? BTW, very much liked chapter 2, best in the book. Louis.

2 3

Compilation errors with using Boost Interval types
by Chidambaram Annamalai 09 Aug '10

09 Aug '10

Hi, I wanted to use the boost::numeric::interval<T> types for simple arithmetic and elementary functions. I expected them to work out of the box (at least for T=double). However, only arithmetic (+,-,*,/) works. I get name look up errors when I try something like sin(..) or cos(..) of an Interval type. The example with the compiler errors are here: http://codepad.org/4LmWiHiS I tried grepping my global include directory for what the compiler reported it couldn't find (cos_up, cos_down) and the only place it seems to be defined in is: [00:57][~]$ grep -iRl cos_up /usr/include/ /usr/include/boost/numeric/interval/transc.hpp And, including that file doesn't seem to help at all as you can see in the paste. Also on a related note, *none* of the examples here: https://gforge.inria.fr/scm/viewvc.php?view=rev&root=std-interval&sortby=da… compile. I expected at least the examples to work out of the box but they don't. Thanks, Chillu

1 0

Stumped!
by Robert Jones 09 Aug '10

09 Aug '10

The code I write needs to compile under several older versions of Boost, as I'm sure is very common. I have been tripped up by using const_pointer_cast, which apparently doesn't exist very far into the past. As far as I can see, implementing this functionality is not trivial, because it's easy to make a mess of the reference counts in shared pointers. I'd like to know when this functionality was introduced, but with the older documentation offline at the moment I'm a bit scuppered. I thought of downloading old versions from source forge, but find there's a gap between 1.33.1 and 1.38 Any ideas how I can find out? Thx - Rob.

4 5

[MultiIndex] is it possible to create multi-key key-extractors?
by Mostafa 09 Aug '10

09 Aug '10

Hi, For the following struct: struct Flower { int type; string name; std::list<string> aliases; }; I want to create a multiindexed container that indexes Flower objects by type, name, and aliases, where aliases is some iteratable collection of strings. I just don't know if it's possible to create a multi-key key-extractor for aliases. So far I've come up with the following: typedef boost::multi_index_container< Flower, indexed_by< ordered_unique< member<Flower, int, &Flower::type> >, ordered_unique< member<Flower, string, &Flower::name> >, ordered_unique< //What goes here ... ? > > > FlowerMngr; Thanks, -Mostafa

4 7

[tuples, multi-index] tuples::get as key extractor
by Igor R 09 Aug '10

09 Aug '10

Hello, Consider the following types: typedef tuple<std::string, int> Item; typedef multi_index_container< Item, indexed_by< ordered_non_unique<global_fun<const Item &, int, &tuples::get<1> > > > > Items; The above types don't compile (MSVC10) because of &tuples::get<1>: C2440: 'specialization' : cannot convert from 'overloaded-function' to 'int (__cdecl *)(boost::tuples::tuple<T0,T1> &)' It seems there're several overloaded function templates "tuples::get"... So, what's the correct type of the appropriate function? Or -- is there a more "robust" way to extract the key (without defining explicit function-object)? Thanks!

2 2

Documentation of older releases?
by Robert Jones 08 Aug '10

08 Aug '10

Let me admit up-front that this ignorance is probably my own fault as I've not been paying attention lately, but I see the main documentation list now only includes releases 1.42 & 1.43. Didn't this used to include releases back to about 1.31 or something? Does this mean that documentation of older releases are no longer hosted on the website? Thx - Rob.

2 1

Re: [Boost-users] VS2010 and /clr: fatal error C1001: An internal error has occurred in the compiler
by Soren Dreijer 08 Aug '10

08 Aug '10

> Probably because that causes some assembly to be emitted because of > the efficiency requirements of the mutex, and the CLR does not support > direct assembly. Either get out of the CLR, or do not use anything > that uses low assembly constructs (CLR has its own thread and mutex > classes anyway). Well, so I can't "just get out of the CLR". As I mentioned earlier, this is a managed C++ project that exposes native functionality from other libraries to .NET. That is, I *have* to include the native headers, which in turn include boost. This worked fine in VS2008, too, so it sounds more like a regression in the compiler I think. Also, when compiling for x86, I see the following warning but everything still compiles (the warning is to be expected, as you say yourself): 4>boost_1_43_0_sdk\include\boost\thread\win32\basic_timed_mutex.hpp(160) : warning C4793: 'boost::detail::basic_timed_mutex::unlock' : function compiled as native : 4> Found an intrinsic not supported in managed code / Soren

2 1

[variant] GNU MP Bignum (gmp) compatibility
by Sven Heithecker 08 Aug '10

08 Aug '10

Hello ! The following code compiles correctly and does what I want: #include <iostream> #include <string> #include <boost/variant.hpp> #include <gmpxx.h> using namespace std; typedef boost::variant<int,string> var_t; // COMMENT //typedef boost::variant<mpz_class,mpq_class> var_t; // UNCOMMENT class add : public boost::static_visitor<var_t> { public: template<typename T> var_t operator()(const T &op1, const T &op2) const { return(op1+op2); } template<typename T1, typename T2> var_t operator()(const T1 &op1, const T2 &op2) const { throw; // just for demonstration } }; int main() { var_t v0(2); var_t v1(3); var_t r0 = apply_visitor(add(),v0,v1); cout << r0 << endl; // "5" var_t v10("Hello, "); var_t v11("World !"); var_t r10 = apply_visitor(add(),v10,v11); cout << r10 << endl; // "Hello, World !" var_t error= apply_visitor(add(),v0,v10); // try to add "2" and "Hello, " cout << error << endl; // never executed } However, if I change to the "gmp" Variant typedef indicated above, compilation fails with an error in the boost::variant sources. I tried to extract the core error message: [...] /usr/include/boost/variant/variant.hpp:1293: error: call of overloaded ‘initialize(void*, const char [8])’ is ambiguous /usr/include/boost/variant/detail/initializer.hpp:89: note: candidates are: static int boost::detail::variant::make_initializer_node::apply<BaseIndexPair, Iterator>::initializer_node::initialize(void*, typename boost::call_traits<typename boost::unwrap_recursive<typename boost::mpl::deref<Iterator>::type>::type>::param_type) [with BaseIndexPair = boost::mpl::pair<boost::detail::variant::make_initializer_node::apply<boost: :mpl::pair<boost::detail::variant::initializer_root, mpl_::int_<0> >, boost::mpl::l_iter<boost::mpl::list2<__gmp_expr<__mpz_struct [1], __mpz_struct [1]>, __gmp_expr<__mpq_struct [1], __mpq_struct [1]> > > >::initializer_node, mpl_::int_<1> >, Iterator = boost::mpl::l_iter<boost::mpl::list1<__gmp_expr<__mpq_struct [1], __mpq_struct [1]> > >] <near match> /usr/include/boost/variant/detail/initializer.hpp:89: note: static int boost::detail::variant::make_initializer_node::apply<BaseIndexPair, Iterator>::initializer_node::initialize(void*, typename boost::call_traits<typename boost::unwrap_recursive<typename boost::mpl::deref<Iterator>::type>::type>::param_type) [with BaseIndexPair = boost::mpl::pair<boost::detail::variant::initializer_root, mpl_::int_<0> >, Iterator = boost::mpl::l_iter<boost::mpl::list2<__gmp_expr<__mpz_struct [1], __mpz_struct [1]>, __gmp_expr<__mpq_struct [1], __mpq_struct [1]> > >] <near match> [...] After having a look at the gmp documentation, especially the "12.6 C++ Interface Limitations" chapter, I think this has something to do with the various conversion functions between the gmp classes and the way how the gmp classes are implemented. Is there any (easy?) way to make boost::variant and gmp compatible ? Regards, Sven -- Sven Heithecker GPG Fingerprint: 527F 06C2 BF51 96CE DFF7 CC89 1A6B DFD5 BEC2 02A6

1 0