February 2013 - Boost-users

[fusion, phoenix, spirit2] C++ orientation
by Michael Powell 26 Feb '13

26 Feb '13

Hello, It's been a little while since I've gone into much depth in C++. Been in the C# world for recent years and have a fairly advanced view of that world, through concepts like push-oriented dependency injection, pull-oriented inversion of control (less preferred), functional style programming using anonymous actions, functions, lambdas, and so on. I am working on a project now where we are adopting C++ into the mix, and I would like to get a much better sense for Spirit (likely Spirit2) for purposes of how to run sequences of complex data through various calculations, finds, counts, etc, probably needing to dynamically parse these calculations. Our data will be a series of spiked tuples (one or more) of n-dimensional packed values, and we'll probably need to dynamically parse calculations on the n-dimensions to determine did we satisfy the solution. I don't know if that makes any sense to anyone. It's all very conceptual, very abstract, at this stage of the game. I know, or have a sense, exactly how I might do it in C#, .NET 4, and so on. A little less so what's available in C++, much less boost, spirit, phoenix, fusion, etc. Any insight where to begin would be helpful. Regards, Michael

10 19

[Regex] MSVC 2008 (VC9) fails to compile regex_raw_buffer.hpp on 1_49
by Larry 26 Feb '13

26 Feb '13

I get an error just including regex.hpp in a program that used to work on earlier VC versions. 1>c:\boost\include\boost-1_49\boost\regex\v4\regex_raw_buffer.hpp(184) : error C2665: 'operator new' : none of the 6 overloads could convert all the argument types 1> c:\program files (x86)\microsoft visual studio 9.0\vc\include\new(57): could be 'void *operator new(size_t,void *) throw()' 1> c:\program files (x86)\microsoft visual studio 9.0\vc\include\new(86): or 'void *operator new(size_t,const std::nothrow_t &) throw()' 1> while trying to match the argument list '(const char [64], long)' 1>c:\boost\include\boost-1_49\boost\regex\v4\perl_matcher_non_recursive.hpp(110) : error C2059: syntax error : '*' inline raw_storage::raw_storage(size_type n) { start = end = static_cast<pointer>(::operator new(n)); <--- offending line for first error BOOST_REGEX_NOEH_ASSERT(start) last = start + n; } struct save_state_init { saved_state** stack; save_state_init(saved_state** base, saved_state** end) : stack(base) { *base = static_cast<saved_state*>(get_mem_block()); *end = reinterpret_cast<saved_state*>(reinterpret_cast<char*>(*base)+BOOST_REGEX_BLOCKSIZE); --(*end); (void) new (*end)saved_state(0); <---- offending line for second error ........... I looked to see if this was a known failure but apparently missed it if it was already covered. It is probably not a Boost problem. The earlier versions were compiled on 32 bit systems and this is a 64 bit system (Win7). I assume I have probably missed some configuration item to make this work. I am fairly certain I did compile this before under VC9 on a 32 bit system (Vista). (Boost was built from source on this system and this compiler.) Any pointers for resolution would be appreciated. Larry

3 4

Boost user's guide or getting started guide?
by John Lilley 25 Feb '13

25 Feb '13

Greetings! I feel that many questions I have about Boost.Geometry are really questions about the "Boost way and true path". Is there a good place to get started with Boost.Geometry specifically or Boost in general? I feel that I am locking in some basic understanding of how all of these templates fit together, what are Ranges, what are "concepts" etc. Where do I RTFM? Thanks, john

4 3

nest mpl::lambda
by Niitsuma Hirotaka 25 Feb '13

25 Feb '13

I would like to use mpl::lambda in nest. In this case works fine. --------- typedef boost::mpl::lambda<boost::mpl::max<boost::mpl::_1,boost::mpl::_2> > my_max; typedef boost::mpl::fold< boost::mpl::vector_c<int,1,7,0,-2,5,-1> , boost::mpl::int_<6> //, boost::mpl::max<boost::mpl::_1,boost::mpl::_2> , my_max >::type r2; BOOST_MPL_ASSERT(( boost::is_same< r2, boost::mpl::integral_c<int, 7> > )); --------- However --------- typedef boost::mpl::lambda< boost::mpl::fold< boost::mpl::_1 ,boost::mpl::front<boost::mpl::_1> ,my_max >::type >::type my_max_types; typedef my_max_types::apply<boost::mpl::vector_c<int,1,7,0,-2,5,-1> >::type r3; BOOST_MPL_ASSERT(( boost::is_same< r3, boost::mpl::integral_c<int, 7> > )); ///error => becomes boost::mpl::integral_c<int, 1> --------- How to define such nest lambda in mpl?

2 6

Re: [Boost-users] [parallel graph] error building parallel BGL
by The Maschine 25 Feb '13

25 Feb '13

Hi ./b2 libs/graph_parallel/example No it didnt work. Tasos ####### "typeinfo for boost::mpi::exception", referenced from: typeinfo for boost::exception_detail::error_info_injector<boost::mpi::exception>in dijkstra_shortest_paths.o "vtable for boost::mpi::exception", referenced from: boost::mpi::exception::exception(boost::mpi::exception const&)in dijkstra_shortest_paths.o NOTE: a missing vtable usually means the first non-inline virtual member function has no definition. ld: symbol(s) not found for architecture x86_64 collect2: ld returned 1 exit status "g++" -L"/usr/local/Cellar/open-mpi/1.6.3/lib" -o "bin.v2/libs/graph_parallel/example/darwin-4.2.1/debug/dijkstra_shortest_paths" "bin.v2/libs/graph_parallel/example/darwin-4.2.1/debug/dijkstra_shortest_paths.o" "bin.v2/libs/system/build/darwin-4.2.1/debug/libboost_system.dylib" "bin.v2/libs/graph_parallel/build/darwin-4.2.1/debug/libboost_graph_parallel.dylib" "bin.v2/libs/serialization/build/darwin-4.2.1/debug/libboost_serialization.dylib" -lm -lmpi -lmpi_cxx -g ...failed darwin.link bin.v2/libs/graph_parallel/example/darwin-4.2.1/debug/dijkstra_shortest_paths... ...failed updating 2 targets...

2 1

Re: [Boost-users] [parallel graph] error building parallel BGL
by The Maschine 25 Feb '13

25 Feb '13

Hi Jeremiah, I can confirm that ParallelBGL that comes with 1.53 compiles. The problem with this one is that when I try to compile the example (again from 1.53) I get this problem: /usr/local/include/boost/graph/distributed/mpi_process_group.hpp:419:28: No viable conversion from 'shared_ptr<boost::graph::distributed::mpi_process_group::impl>' to 'bool' 1) Any ideas? 2) Any location that I can find more examples of the Parallel BGL? Best, Tasos

3 3

Re: [Boost-users] [parallel graph] error building parallel BGL
by The Maschine 25 Feb '13

25 Feb '13

Hi, I used the BFS example from "lib/graph_parallel/example/" Maybe Im not linking the right files? Thanks, Tasos

2 1

MSVC Visualizer for Fusion Vectors
by Rodrigo Madera 25 Feb '13

25 Feb '13

Are there any plans (or actual implementations) for Visual Studio debug visualizers for Fusion Vectors? Thank you, Rodrigo

2 2

Re: [Boost-users] Boost-users Digest, Vol 3370, Issue 2
by Brammert Ottens 25 Feb '13

25 Feb '13

> Brammert Ottens > > Hi, > > I've had a closer look and I was too quick in dismissing your solution. however, the on_label_popped(cur_label) gets a const reference to the label. I've used a const_cast to remove the const (see code below), but now the code crashes when cleaning up the labels using deallocate. Any ideas on what could be wrong? > > template<class Label, class Graph> > void QRCSPPVisitor::on_label_popped( const Label& label, const Graph& graph ) > { > RCSPPVisitorData().IncrementLabelsPopped(); > > if ( TimedOut() || GetMillis(StartTime()) > MaxMillis() ) > { > this->SetTimedOut(); > const_cast<Label&>(label).b_is_dominated = true; > } > } I think you can just change the parameter declaration to be a non-const reference. Can you please check where the crash occurs and whatever information you can get about the cause? Thanks. -- Jeremiah Willcock Hi, The crash occurs in the final loop, where you are going over vec_vertex_labels. It crashes when calling l_alloc.deallocate( (*csi).get(), 1 );. We use the standard std::allocator. Unfortunately the code has been encapsulated into a much larger project, and the only information that I have is what windows gives me: Problem Event Name: APPCRASH Application Name: QServer_D64.exe Application Version: 2013.6.6145.0 Application Timestamp: 512b792f Fault Module Name: KERNELBASE.dll Fault Module Version: 6.1.7601.18015 Fault Module Timestamp: 50b8479b Exception Code: 40010006 Exception Offset: 0000000000009e5d OS Version: 6.1.7601.2.1.0.256.1 Locale ID: 1043 Additional Information 1: 02d5 Additional Information 2: 02d5a24c112729571ddd0daf5f60defb Additional Information 3: 563e Additional Information 4: 563e19c789cadcf1ddb48080eecb0875 Removing the const from Label doesn't help, the code still crashes. So if you can put a visitor that allows to jump out of the loop prematurely that would be immensely helpful! Thanks already for the quick help. Brammert

1 0

Re: [Boost-users] Boost-users Digest, Vol 3369, Issue 2
by John M. Dlugosz 25 Feb '13

25 Feb '13

On 2/24/2013 12:44 PM, boost-users-request-at-lists.boost.org |Boost/Allow to home| wrote: > Are you implying this of Boost libraries in general ? If so I beg to > differ. Boost libraries cover more than .Net libraries do in the realm > of general purpose programming. The .Net API very nicely covers APIs > specific to Windows programming and some general purpose APIs, but not > nearly so much as Boost does of the latter. > > I do not mean to start an opinion war but I think your reply above gives > the OP the wrong idea about the strengths of Boost libraries, especially > as he is new to Boost. And yes I have programmed in .Net ( both C# and > C++/CLI ) pretty extensively. <http://channel9.msdn.com/Events/GoingNative/GoingNative-2012/Interactive-Pa…> at about 30 Minutes, “no large libraries [for C++]” <http://channel9.msdn.com/Events/GoingNative/GoingNative-2012/C-11-VC-11-and…> Thesis starts at 1:15, lasts till nearly the end. Now I would agree that his counting picks up a lot of "bloat" from API wrappers, but he does mention more specific examples of libraries that are needed for standard C++. Could you elaborate on your remarks, using this as a starting point of where we are coming from? —John

1 0