October 2005 - Boost-users - lists.boost.org

Boost T-Shirts @ Cafe Press
by Jeff Garland 27 Oct '05

27 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

[ptr_container] warnings about shadowed members
by Samuel Landau 27 Oct '05

27 Oct '05

Hello, I am trying Boost 1.33 ptr_container with GCC 4.0.1 on Linux. When including a ptr_container header (say, ptr_vector.hpp), compiling code with the "-Wshadow" flag, some warnings are issued about shadowed members. I am wondering if this might be a problem, even if my snippet code compiles and seems to run fine so far for me. output example (with boost/ in current directory) : $ cat main.cpp #include <boost/ptr_container/ptr_vector.hpp> int main() {} $ g++ -I. -Wshadow main.cpp ./boost/ptr_container/detail/move.hpp: In constructor 'boost::ptr_container_detail::move_ptrs::move_source<Ptr>::move_source(Ptr&)': ./boost/ptr_container/detail/move.hpp:18: warning: declaration of 'ptr' shadows a member of 'this' ./boost/ptr_container/detail/static_move_ptr.hpp: In member function 'void boost::ptr_container_detail::static_move_ptr<T, Deleter>::check(const +boost::ptr_container_detail::static_move_ptr<TT, DD>&)': ./boost/ptr_container/detail/static_move_ptr.hpp:156: warning: declaration of 'ptr' shadows a member of 'this' ./boost/ptr_container/exception.hpp: In constructor 'boost::bad_ptr_container_operation::bad_ptr_container_operation(const char*)': ./boost/ptr_container/exception.hpp:27: warning: declaration of 'what' shadows a member of 'this' ./boost/ptr_container/exception.hpp: In constructor 'boost::bad_index::bad_index(const char*)': ./boost/ptr_container/exception.hpp:41: warning: declaration of 'what' shadows a member of 'this' ./boost/ptr_container/detail/reversible_ptr_container.hpp: In member function 'ForwardIterator +boost::ptr_container_detail::reversible_ptr_container<Config, CloneAllocator>::advance(ForwardIterator, typename Config::void_container_type::size_type)': ./boost/ptr_container/detail/reversible_ptr_container.hpp:287: warning: declaration of 'begin' shadows a member of 'this' -- Samuel Landau

2 1

[boost-users] [MPL.Math] gcd_impl improved
by Cromwell Enage 27 Oct '05

27 Oct '05

The gcd_impl specialization for integral constants is now based on the binary GCD algorithm. BOOST_STATIC_CONSTANT is now used over typedef at most intermediate steps. You should see much faster compile times, at least for the gcd_lcm test and example programs. Cromwell D. Enage __________________________________ Yahoo! FareChase: Search multiple travel sites in one click. http://farechase.yahoo.com

1 0

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

>Is there a page somewhere that describes the trick, just out of interest >like. C++ Template Metaprogramming, page 205 Modern C++ Design, somewhere AFAIK - Barton & Nackmann trick is used for class composition: template< typename Derived > struct Add { int add() { // B & N-trick: access to the Composed class D & d = static_cast< Derived & >( * this); return d.first() + d.second(); } }; struct First { int first() { return 1; } }; struct Second { int second() { return 2; } }; class Composed : inherit_scattered< vector< First, Second, Add< Composed > > > {}; void main() { Composed c; Std::cout << c.add() << std::endl; }

2 1