- Boost-users - lists.boost.org

A basic interior properties question
by Gat, Yoram 28 Oct '05

28 Oct '05

Hi all, I just started using BGL and created the following program (MSVS 7.1), storing some data in an interior property and then retrieving them: #include "boost/graph/adjacency_list.hpp" struct NodeProperties { int a, b, c; }; typedef boost::adjacency_list< boost::listS,boost::listS,boost::undirectedS, boost::property<boost::vertex_all_t,NodeProperties> > Graph; typedef boost::graph_traits<Graph>::vertex_descriptor VertexDescriptor; typedef boost::property_map<Graph,boost::vertex_all_t>::type AllMap; int main(int argc, char const *argv[]) { Graph graph; AllMap allMap = get(boost::vertex_all,graph); VertexDescriptor vd = boost::add_vertex(graph); NodeProperties np; np.a = 1; np.b = 2; np.c = 3; allMap[vd] = np; NodeProperties np1 = allMap[vd]; // const AllMap::value_type &vt = boost::get(allMap,vd); // NodeProperties np1 = vt.m_value; return 0; } For which I get the following error: c:\Documents and Settings\ygat\Desktop\Work\InternalProperty\InternalProperty.cpp(24): error C2440: 'initializing' : cannot convert from 'boost::detail::vertex_property_map<Graph,PropertyTag>::Property' to 'NodeProperties' with [ Graph=Graph, PropertyTag=boost::vertex_all_t ] After some poking around, I discovered that if I replace the offending line with the two commented lines, things work. Is this the right way to retrieve the data? Why can the NodeProperties object be on the rhs but not on the lhs of an assignment to the property map? Thanks, Yoram

1 0

[multi_index] non-intrusive keys
by Adi Shavit 28 Oct '05

28 Oct '05

Hi, In my application I used a multi_map of multi_maps as a data structure to provide indexing with 2 keys, say long strings. Access is logarithmic on both keys and the keys are not part of the stored object, thus are potentially only stored once. I'd like to use multi_index as a more elegant replacement. However, I'm not clear about where the keys are stored. In my app, the keys are stored separately from the data, as the keys to maps, as part of the map data structure. It seems that in multi_index the keys have to be inside the stored data. In this case, 10 objects with the same key(s) will all have to hold copies of the keys (as members). Is there a way to avoid this? I hope I'm clear, Thanks, Adi

1 0

singleton in boost::library?
by Oliver.Kowalke＠infineon.com 28 Oct '05

28 Oct '05

Hello, why no singleton-template is available in boost?

3 2

Boost T-Shirts @ Cafe Press
by Jeff Garland 28 Oct '05

28 Oct '05

Hi All - As everyone is probably already aware, we have a new Boost logo that was created by booster Zoltan 'cad' Juhasz. As part of the final logo refinement process, the moderators asked Zoltan if he could create T-Shirts with the new logo. We also felt it was appropriate that Zoltan be the beneficiary of any proceeds associated with the T-Shirts since it was his tireless work that went into the logo and T-Shirt design. With that in mind, I'm delighted to announce that Zoltan worked out some T-Shirt designs and has set up a couple shops at Cafe Press so that people can purchase Boost T-Shirts. - http://www.cafepress.com/boostlibs - http://www.cafepress.com/boostlibsb If you have suggestions or other ideas you can find Zoltan's contact info at: http://people.inf.elte.hu/cad Jeff

1 0

result_of && VC7.1
by Richard Crossley 27 Oct '05

27 Oct '05

Hi, Experimenting with boost::result_of under VC7.1 and 8... --- template <typename T> struct A{ template <typename F> struct result; template <typename U,typename V> struct result<A(U,V)>{ typedef U type; }; template <typename U,typename V> U operator()(U const&,V const&) { return U(); } }; //typedef A<int>::result<A<int>(int,int)>::type result_; typedef boost::result_of<A<int>(float,char)>::type result; --- c:\boost\boost\utility\result_of.hpp(43) : error C2027: use of undefined type 'A<T>::result<F>' with [ T=int ] and [ F=A<int> (float,char) ] Am I actually using result_of correctly here, or have I misunderstood something? If the result_ typedef is uncommented then it behaves as I would expect. Is there a way to successfully use result_of like this with VC? Thanks, Richard.

1 0

SUN Compiler Feasibility
by Frank Boyle 27 Oct '05

27 Oct '05

Hi Folks, I'm using the sun compiler: CC: Sun WorkShop 6 update 2 C++ 5.3 Patch 111685-15 2003/05/14 I need to generate a high performance logging process and immediately thought of the boost filesystem service. I have downloaded boost_1_33_0. However, I'm getting a lot of compiler problems using the templates which I've since realised are common when using this compiler. i.e. errors relating to boost::iterator_core_access::equal and copious warnings: "/product/3rd_Party/boost/boost_1_33_0/boost/iterator/detail/facade_iterator_category.hpp", line 189: Where: Specialized in boost::detail::facade_iterator_category_impl<boost::bidirectional_traversal_tag, const _STL::basic_string<char, _STL::char_traits<char>, _STL::allocator<char>>, const _STL::basic_string<char, _STL::char_traits<char>, _STL::allocator<char>>&>. "/product/3rd_Party/boost/boost_1_33_0/boost/mpl/eval_if.hpp", line 41: Where: Specialized in boost::mpl::eval_if<boost::detail::is_iterator_category<boost::bidirectional_traversal_tag>, boost::mpl::identity<boost::bidirectional_traversal_tag>, boost::detail::facade_iterator_category_impl<boost::bidirectional_traversal_tag, const _STL::basic_string<char, _STL::char_traits<char>, _STL::allocator<char>>, const _STL::basic_string<char, _STL::char_traits<char>, _STL::allocator<char>>&>>. "/product/cs_common/3rd_Party/boost/boost_1_33_0/boost/iterator/detail/facade_iterator_category.hpp", line 202: Where: Specialized in boost::detail::facade_iterator_category<boost::bidirectional_traversal_tag, const _STL::basic_string<char, _STL::char_traits<char>, _STL::allocator<char>>, const _STL::basic_string<char, _STL::char_traits<char>, _STL::allocator<char>>&>. My question is quite simple: Is it feasible for me to use boost's filesystem utility with this version of the Sun compiler? Should I persist? Many Thanks in advance, Joe ___________________________________________________________ To help you stay safe and secure online, we've developed the all new Yahoo! Security Centre. http://uk.security.yahoo.com

2 1

storing templates in boost::mpl::vector instead of complete types
by Oliver.Kowalke＠infineon.com 27 Oct '05

27 Oct '05

Hello, is it possible to store templates instead of complete types in boost::mpl::vector? I want to compose a class out of templates: template< typename B, typename E, typename Comp > struct Scattered; template< typename B, typename E, typename Comp > struct Scattered : public mpl::deref< B >::type< Comp >, // <<== inherit from the template, Comp == Special public Scattered< typename mpl::next< B >::type, E, Comp > {}; template< typename E, typename Comp > struct Scattered< E, E, Comp > {}; template< typename Sequence, typename Comp > struct InheritScattered : public detail::Scattered< typename mpl::begin< Sequence >::type, typename mpl::end< Sequence >::type, Comp > {}; template< typename D > class ReadN { void readn() { D & d = static_cast< D & >( * this); d.some_function(); .... } }; template< typename D > class WriteN { void writen() { D & d = static_cast< D & >( * this); ... } }; temlate< typename TL > struct Basic {}; typedef BaiscX< mpl::vector< ReadN, WriteN > > Special; // <<== composing SpecialX (inherit from ReadN which is a template) Special s; s.readn(); s.writen();

2 1

Re: [Boost-users] multi-inheritance from elements of mpl::vector?==> inherit_scattered ( for Barton & Nackmann trick)
by Oliver.Kowalke＠infineon.com 27 Oct '05

27 Oct '05

>>no reply - so I found a solution by myself (used for the Barton & >> Nackmann trick)! >> template< typename D > >> struct F >> { >> std::string f() >> { return static_cast< D & >( * this).str_; } >> }; >> >> template< typename D > >> struct G >> { >> std::string g() >> { return static_cast< D & >( * this).str_; } >> }; > >That's not the B&N trick; it's simple CRTP. >The B&N trick is the one where you stick a friend function that >applies to the derived class in the base class. See the operators >library for reference. Sorry - in your book (C++ Template Metaprogramming, page 205) I can read: "The technique of using a static_cast with CRTP to reach the derived object is sometimes called the "Barton and Nackmann trick" ...". Is this wrong? So long, Oliver

3 2

filesystem library with SparcWorks compiler?
by Chris Newbold 27 Oct '05

27 Oct '05

It appears that the filesystem library and the SparcWorks compiler have some disagreements. I'm running into the same issues as reported on the regression test page here (an example of which is here: http://tinyurl.com/ey5ld) What I'm wondering is if there are any potential patches to Boost to address these problems or if there is some effort to do so. Alternatively, is the SparcWorks compiler just too busted to deal with this library? -Chris

1 0

ptr_vector problem when storing it inside a map
by Oleg Smolsky 27 Oct '05

27 Oct '05

Hi all, I've just started looking at ptr_vector<> and got stuck on the following issue: putting a boost::ptr_vector<Simple> into a map doesn't compile. See the following test case: class Simple { public: virtual ~Simple() { } void foo() { doFoo(); } protected: virtual void doFoo() { } }; void Test() { std::map<std::string, boost::ptr_vector<Simple> > map; map["key"].push_back(new Simple); } I'm using VS2005 RC1 (8.0.50727.26) and the compiler output is as follows: 1>c:\work\3rd-party\boost\include\boost-1_33\boost/ptr_container/ptr_sequence_adapter.hpp(595) : error C2248: 'boost::ptr_container_detail::reversible_ptr_container<Config,CloneAllocator>::reversible_ptr_container' : cannot access private member declared in class 'boost::ptr_container_detail::reversible_ptr_container<Config,CloneAllocator>' 1> with 1> [ 1> Config=boost::ptr_container_detail::sequence_config<Simple,std::vector<void *,std::allocator<void *>>>, 1> CloneAllocator=boost::heap_clone_allocator 1> ] 1> c:\work\3rd-party\boost\include\boost-1_33\boost/ptr_container/detail/reversible_ptr_container.hpp(294) : see declaration of 'boost::ptr_container_detail::reversible_ptr_container<Config,CloneAllocator>::reversible_ptr_container' 1> with 1> [ 1> Config=boost::ptr_container_detail::sequence_config<Simple,std::vector<void *,std::allocator<void *>>>, 1> CloneAllocator=boost::heap_clone_allocator 1> ] 1> This diagnostic occurred in the compiler generated function 'boost::ptr_sequence_adapter<T,VoidPtrSeq,CloneAllocator>::ptr_sequence_adapter(const boost::ptr_sequence_adapter<T,VoidPtrSeq,CloneAllocator> &)' 1> with 1> [ 1> T=Simple, 1> VoidPtrSeq=std::vector<void *,std::allocator<void *>>, 1> CloneAllocator=boost::heap_clone_allocator 1> ] Can anybody shed some light on the subject? Thanks, Oleg.

3 4