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

Test Tools Linking Problem
by Merrill Cornish 26 Jul '05

26 Jul '05

I am an experienced programmer, but new to the Boost Libraries. I am using Boost 1.32.0 with MinGW under XP Pro. I have successfully compiled and link a program to the boost_regex-mgw-mt-1_32.dll Now I am trying use Boost test tools to write tests for it. When I compile and link, the compile passes without warnings, but the link fails with undefined reference to 'boost::test_tools::tt_detail::message_impl( boost::basic_wrap_stringstream<char>&, boost::unit_test::basic_cstring<char const>, unsigned int)' I initially set up to statically link with libboost_unit_test_framework-mgw-mt-1_32.lib. When I got the undefined reference, I added libboost_test_exec_monitor-mgw-mt-1_32.lib and libboost_prg_exec_monitor-mgw-mt-1_32.lib, "just in case." Still undefined. Eventually, I tried added in almost all of the static libraries PLUS copies all of the DLL into the program directory. Still undefined. At that point, I figured that something more than an incomplete makefile is the problem here. Anyone have any idea what I've missing? Merrill

2 1

Re: [Boost-users] minimal spanning trees & Visual C++ 7.1
by Jasper, Alexander 26 Jul '05

26 Jul '05

> -----Ursprüngliche Nachricht----- > Von: boost-users-bounces(a)lists.boost.org > [mailto:boost-users-bounces@lists.boost.org]Im Auftrag von > Felipe Magno > de Almeida > Gesendet: Montag, 25. Juli 2005 22:04 > An: boost-users(a)lists.boost.org > Betreff: Re: [Boost-users] minimal spanning trees & Visual C++ 7.1 > > > On 7/25/05, Amit K. Jain <omlette(a)gmail.com> wrote: > > Thank you, it worked! That sounds good! > > > > It is bizarre behavior on the compiler's part, I hope they get it > > fixed eventually. > > > > -Amit > > > > > Hi! > > > > > > I suffered from ICEs using boost as well. They occured > using multi_index lib, though. Turning of "Minimal Rebuild" > helped for me. > > Minimal rebuild has given a lot of headache to me already. It has > introduced bugs in my applications when using boost(I think it > probably gets mistaken with its compilation depedancy and left some > files without compile, that in fact should have been recompiled... I > dont know...). Everyone seems to have the same problems. Often enough the strangest behavior has gone vanished after doing a "rebuild all". Some times it is saying "no relevant changes detected" ... in fact they weren't quite so irrelevant. Have you noticed this annoying bug of VC 7.1 debugger: create 2 (or more) files with the same name in different directories and build static libs from these directories. Put all projects in one solution file. The debugger is unable to distinguish between the 2 files. Every break is magically moved from one file to the other, when stepping code, the cursor moves in one file, but the code is executed in the other file. Annoying! I hope they fixed that with 8.0. > > > > > > > kind regards > > > > > > Alex > > > > _______________________________________________ > > Boost-users mailing list > > Boost-users(a)lists.boost.org > > http://lists.boost.org/mailman/listinfo.cgi/boost-users > > > > > -- > Felipe Magno de Almeida > Developer from synergy and Computer Science student from State > University of Campinas(UNICAMP). > Unicamp: http://www.ic.unicamp.br > Synergy: http://www.synergy.com.br > "There is no dark side of the moon really. Matter of fact > it's all dark." > > _______________________________________________ > Boost-users mailing list > Boost-users(a)lists.boost.org > http://lists.boost.org/mailman/listinfo.cgi/boost-users >

1 0

SImple Boost Lambda problem using arguement as object.
by Thaddeus Olczyk 26 Jul '05

26 Jul '05

This piece of sample code demonstrates my problem quite well: #include <boost/lambda/lambda.hpp> #include <vector> #include <string> #include <iostream> using boost::lambda::_1; class A { std::string s; public: A(std::string sIn):s(sIn){} void print_A() { std::cout<<"A="<<s<<std::endl; } }; int main() { std::string in; std::vector<A *> alist; while(in!="end") { in=""; std::getline(std::cin,in); alist.push_back(new A(in)); } std::for_each(alist.begin(),alist.end(), _1->print_A()); // This is line 30 return 1; } ---- Using g++ ( 3.??) and the latest boost, I get this error: main.cpp: In function `int main()': main.cpp:30: error: base operand of `->' has non-pointer type `const boost::lambda::lambda_functor<boost::lambda::placeholder<1> >' Can anyone explain how to fix this? Thanks -- Thad

3 2

Overload boost::program_options validate() for double?
by Joseph Turian 26 Jul '05

26 Jul '05

Hi, I am having difficulty overloading validate() of boost::program_options to do conversion of doubles. I define it like in the example code ( http://www.boost.org/doc/html/program_options/howto.html#id549525) void validate(boost::any& v, const vector<string>& values, double* target_type, int) { cerr << "TEST\n"; } However, this method is bypassed and the original validate method is used instead. How can I get it to use *my* validate implementation for the double type? Thanks, Joseph -- http://www.cs.nyu.edu/~turian/

1 0

shared_array double free problem
by Jack Nguyen 26 Jul '05

26 Jul '05

Here is the compilable code: #include <boost/shared_array.hpp> #include <iostream> template <typename T> class VectorView; template <typename T> inline std::ostream& operator << (std::ostream& os, const VectorView<T>& VOut) { os<<VOut.size()<<std::endl; for (int i=0; i<VOut.size() && os.good(); ++i){ os<<VOut(i)<<""; os<<std::endl; } return os; } template<typename T> class VectorView { private: int _size; boost::shared_array<T> _vp; public: VectorView(T* ptr, int size) { _size=size; _vp=boost::shared_array<T>(ptr); } T operator () (int pos) const { return _vp[pos]; } int size() const { return _size; } ~VectorView() { } }; int main (void) { double a[]={1,2,3,4,5,6,7}; VectorView<double>V(a,4); std::cout<<V; return 0; } Any workaround?

2 1

Two Threads sharing a mutex
by Christian Henning 26 Jul '05

26 Jul '05

Hi there, please help me out. Seems I lost my brain over the last weekend. ;-)) I'm trying to get myself familiar with the boost::thread lib, by reading an article by Bill Kempf published in the C++ magazine in 2002. Here in listing no. 4 he is using two thread (a reader and a writer ) to access a buffer. In the put() and get() he is using one mutex. If the buffer is empty he uses a condition variable to block the mutex. I don't understand how a writer can then enter the put() and write an item into the buffer? The mutex was just blocked by the reader thread? I guess my question is more fundamental to threads. But hopefully there is someone that can help me out. Thanks ahead, Christian

3 3

BGL Question/Advice which algorithm to use
by Elisha Berns 26 Jul '05

26 Jul '05

Hi, This is a request for some advice/recommendations regarding which algorithm(s) to use with Boost.Graph to solve a graph sorting problem for graphs that are *almost, but not quite* DAGs. If anybody has some insight into this type of situation help will be greatly appreciated. Ideally I do a topological sort to find the most dependent vertex (if that makes sense) which then becomes the root of an n-ary tree, and then build the tree from that vertex recursively finding the out-edges of child vertices. All that works provided the graph is actually a DAG. But sometimes it isn't because there can be small cycles(?) inside the graph where one vertex refers to its parent's parent's vertex. So everything in the graph can be suitable for a DAG except for this small set of vertices that becomes self-referential. This small set of vertices, while not expendable, is certainly never going to contain the desired vertex to create a root, for various reasons. So during the search phase they can be safely ignored, at least that's a fairly safe operating assumption as of now. I still need to find the most dependent vertex, and I can't use topological sort here, so what algorithm can I use? Or if I still need to use topological sort here, would I clone the graph object and remove this set of vertices to create a DAG for the sake of the sort? Many thanks, Elisha Berns e.berns(a)computer.org tel. (310) 556 - 8332 fax (310) 556 - 2839

3 11

problem using boost::filesystem
by Mesut Büyüktokatli 26 Jul '05

26 Jul '05

Hello... I am triing to use the boost::filesystem library. But actually i am stucked using it. I have WindowsXP without a service pack. I am using Visual Studio 6.0 installed in default folders for VS6. I am using stlport but without the iostreams. After uncompressing boost, i compiled boost with bjam with following statement: bjam "-sTOOLS=msvc-stlport" "-sSTLPORT_VERSION=4.6" "-sSTLPORT_PATH=..\stlport" "-sBUILD=debug <stlport-iostream>off" stage Compilation has no errors and all libs are build. Now i want to use boost::filesystem::path, example: #include <boost/filesystem/path.hpp> int main(int argc, char* argv[]) { boost::filesystem::path( "foo" ); } What happens, is that the program stops on that command and does nothing! If i use following code: boost::filesystem::path( "C:\\" ); System makes an kernal32 exception and crashs. Same happens when i use boost::filesystem::exists Could you please help me why that happens??? Thx Mesut

3 4

Re: [Boost-users] performance issue with graph edge removal
by Feldman, Yulik 25 Jul '05

25 Jul '05

I've also noticed that when in/out edge lists are stored in an associative container and the parallel edges are allowed, i.e multiset is used, the functions remove_edge()/clear_vertex() do not work as efficient as they could. The calls to these functions (in my case) result in the following sequence of calls: // O(E/V) template <class EdgeList, class vertex_descriptor> void erase_from_incidence_list(EdgeList& el, vertex_descriptor v, allow_parallel_edge_tag) { boost::graph_detail::erase_if(el, detail::target_is<vertex_descriptor>(v)); } Then: template <class Container, class Predicate> void erase_if(Container& c, Predicate p) { erase_if_dispatch(c, p, container_category(c), iterator_stability(c)); } Then: template <class AssociativeContainer, class Predicate> void erase_if_dispatch(AssociativeContainer& c, Predicate p, associative_container_tag, stable_tag) { typename AssociativeContainer::iterator i, next; for (i = next = c.begin(); next != c.end(); i = next) { ++next; if (p(*i)) c.erase(i); } } The complexity of the search operation is linear, so multiset doesn't provide a performance advantage over containers such as vector or set. However, the function erase_if_dispatch() could use multiset's equal_range() member function, which would result in O(log n) complexity of the search. I understand that the reason erase_if_dispatch() is written in the way it is written, is probably that the function is generic with regard to the predicate and doesn't do assumptions on it. However, in the case erase_if_dispatch() is called from erase_from_incidence_list(), where the predicate is set, the function (its special specialization) could make assumption that the predicate is actually the same as the predicate used to sort the in/out edge list itself (the target vertex), and therefore could use equal_range() to boost the performance of the function. --Yulik.

2 1

Re: [Boost-users] minimal spanning trees & Visual C++ 7.1
by Amit K. Jain 25 Jul '05

25 Jul '05

Thank you, it worked! It is bizarre behavior on the compiler's part, I hope they get it fixed eventually. -Amit > Hi! > > I suffered from ICEs using boost as well. They occured using multi_index lib, though. Turning of "Minimal Rebuild" helped for me. > > kind regards > > Alex

2 1