- Boost-users - lists.preview.boost.org

shared_ptr memory leak (newb)
by Aaron Campbell 07 Aug '03

07 Aug '03

why does this leak memory from the first call to "new"? how can I prevent the leak? wchar_t* pmem = new wchar_t[10]; boost::shared_ptr<std::wstring> cp ( new std::wstring(pmem) ); thanks for the help, .Aaron

1 0

Problems with if_then_else_return in boost::lambda library
by Ioffe, Robert 07 Aug '03

07 Aug '03

Hi, If I have a vector of strings vector<string> v; v.push_back("One"); v.push_back("Two"); v.push_back("Three"); v.push_back("Four"); v.push_back("Five"); and define arithmetic action plus like this: namespace boost { namespace lambda { struct plain_return_type_2<arithmetic_action<plus_action>, std::string, std::string> { typedef std::string type; }; } } then the following code works just fine: cout << accumulate(v.begin()+1, v.end(), v.front(), _1 + string(", ") + _2) << endl; and the following also works wonderfully: int i = 1; cout << accumulate(v.begin()+1, v.end(), v.front(), _1 + bind(constructor<string>(), var(i)++, ' ') + _2) << endl; Now, if I want to use if_then_else_return as shown below, it fails miserably, producing a huge error message: cout << accumulate(v.begin(), v.end(), string(), if_then_else_return(_1 == string(), _1, _1 + bind(constructor<string>(), var(i)++, ' ') + _2)) << endl; Now for the questions: What is the proper way to use if_then_else_return? Why do I need to define a separate arithmetic action for strings, even though they are a common library type? Is there a way to say the most intuitive: cout << accumulate(v.begin()+1, v.end(), v.front(), _1 + ", " + _2) << endl; Is there way to do more intuitive: cout << accumulate(v.begin()+1, v.end(), v.front(), _1 + string(i++, ' ') + _2) << endl; Thank you for your help. Robert Ioffe [Non-text portions of this message have been removed]

1 0

[function] How to extract a tuple<A0,A1> of arguments from a boost::function<>?
by Fernando Cacciola 07 Aug '03

07 Aug '03

Short question: As the Subject says, given a boost::function{1..N}<>, how can I extract a tuple of all its arguments? Long question: Anyway, what I'm really trying to do is this: Given some classes with different methods: struct X { void f(int n ) ; } ; struct Y { void g(char c, double d ) ; } ; I want to easly create _polymorphic_ closures binding member functions of X /Y with instances of X/Y and function arguments. The goal is the have a deferred 'action' list for easy backtracing of an algorithm. That is, instead of directly calling: X x ; Y y ; x.f(1); y.g('a',1.2); I want to do something that would look like: X x; Y y; ActionList actions; actions.push_back( make_closure(&X::f, x,1) ); actions.push_back( make_closure(&Y::g, y,'a',1.2) ); std::for_each(actions.begin(),actions.end(),execute_action()); TIA Fernando Cacciola

2 2

filesystem compilation on tru64 unix
by allysonlurena 07 Aug '03

07 Aug '03

Hi all, I have been trying to build the filesystem library for one of our alpha servers running tru64 unix. I have managed to compile bjam using the tru64cxx toolset. Then, I went into the filesystem build directory and ran the following, but it has been hanging for over 30 minutes now on the last command you see below. Can anyone help? It could just be our machines, but I am at a loss. Does anyone have any ideas? Any help would be appreciated! Thanks Allyson $ cxx -V Compaq C++ V6.5-014 for Compaq Tru64 UNIX V5.1A (Rev. 1885) Compiler Driver V6.5-014 (cxx) cxx Driver rum:/.../boost_1_30_0/libs/filesystem/build $ bjam "-sTOOLS=tru64cxx65" ...found 432 targets... ...using 1 temp target... ...updating 9 targets... tru64cxx65-C++-action ../../../libs/filesystem/build/bin/libboost_filesystem.a/tru64cxx65/debug/exception.o tru64cxx65-C++-action ../../../libs/filesystem/build/bin/libboost_filesystem.a/tru64cxx65/debug/operations_posix_windows.o cxx: Warning: ../src/operations_posix_windows.cpp, line 236: #767-D conversion from pointer to smaller integer assert( m_imp.get() ); // fails if dereference end iterator ------^ cxx: Warning: ../src/operations_posix_windows.cpp, line 242: #767-D conversion from pointer to smaller integer assert( m_imp.get() ); // fails on increment end iterator ------^ tru64cxx65-C++-action ../../../libs/filesystem/build/bin/libboost_filesystem.a/tru64cxx65/debug/path_posix_windows.o tru64cxx65-C++-action ../../../libs/filesystem/build/bin/libboost_filesystem.a/tru64cxx65/debug/convenience.o tru64cxx65-Archive-action ../../../libs/filesystem/build/bin/libboost_filesystem.a/tru64cxx65/debug/libboost_filesystem.a ar: Warning: creating ../../../libs/filesystem/build/bin/libboost_filesystem.a/tru64cxx65/debug/libboost_filesystem.a Ranlib ../../../libs/filesystem/build/bin/libboost_filesystem.a/tru64cxx65/debug/libboost_filesystem.a ...using <libs!filesystem!build/libboost_filesystem.a/tru64cxx65/release>exception.o... tru64cxx65-C++-action ../../../libs/filesystem/build/bin/libboost_filesystem.a/tru64cxx65/release/operations_posix_windows.o So, it seems to hang at the last command above. Thanks again! Allyson

3 2

cvs checkout probs
by Angus Leeming 07 Aug '03

07 Aug '03

Hi, I have just tried to checkout the boost cvs tree from sourceforge, but get stuck at cvs server: Updating boost/libs/numeric/mtl/test cvs server: [05:07:19] waiting for anoncvs_boost's lock in /cvsroot/boost/boost/libs/numeric/mtl/test I can subsequently check out the remaining stuff with $ cat tmp pool preprocessor property_map python random rational regex signals smart_ptr spirit static_assert test thread timer tokenizer tuple type_traits utility $ for dir in `cat tmp`; do cvs -z3 co boost/libs/$dir; done so it appears that the problem is specific to the boost/libs/numeric/mtl/test directory. Kind regards, Angus

1 0

[boost] filesystem lib & linking (MSVC.NET)
by Jonathan Roewen 07 Aug '03

07 Aug '03

Hi All, I'm trying to use the filesystem lib, but I'm having some problems figuring out how to build it properly. I'm using MSVC.NET on WindowsXP, and I ran the bjam tool like in the docs for the whole set of libraries. It took me a while to figure out where it had put everything, and finally found a file called libboost_filesystem.lib. I copied it to a convenient location, added the library path and lib file to my project settings, and then tried to link my app. MSVC then gave a bunch of linker errors, and suggested to add MSVCRT to the nodefaultlib setting, so I did that, and got less linker errors. All that is left is some thing to do with exception & string classes. Linker errors are all about multiply defined symbols. I suppose I'm doing something wrong somewhere along the line, but not sure what. I can't tell if bjam is actually doing what it's supposed to, and the documentation around the place can be pretty sparse in places, so I hope someone can point out what I can do to get things working. Regards, Jonathan Roewen

4 4

Re: what is wrong with boost::bind?
by Jason Shirk 06 Aug '03

06 Aug '03

"yxh" <yg-boost-users(a)m.gmane.org> wrote in message news:<bgqfkv$eeu$1(a)main.gmane.org>... > I solved it by using Visual Assistant .NET I can't seem to reproduce the problem you have reported. If you could send me (jasonsh(a)microsoft.com) a project that reproduces the problem, I'd appreciate it. You might consider trying Intellisense a bit more. Because we use a variant of our C++ compiler for Intellisense, it has capabilities that are beyond many other editors. It's far from perfect, as you've found, but as far as we are aware, the template support for auto completion is much better than with any other similar tool. -- Jason Shirk VC++ Compiler Team

1 0

Problems with if_then_else_return in boost::lambda library
by Ioffe, Robert 06 Aug '03

06 Aug '03

Hi, If I have a vector of strings vector<string> v; v.push_back("One"); v.push_back("Two"); v.push_back("Three"); v.push_back("Four"); v.push_back("Five"); and define arithmetic action plus like this: namespace boost { namespace lambda { struct plain_return_type_2<arithmetic_action<plus_action>, std::string, std::string> { typedef std::string type; }; } } then the following code works just fine: cout << accumulate(v.begin()+1, v.end(), v.front(), _1 + string(", ") + _2) << endl; and the following also works wonderfully: int i = 1; cout << accumulate(v.begin()+1, v.end(), v.front(), _1 + bind(constructor<string>(), var(i)++, ' ') + _2) << endl; Now, if I want to use if_then_else_return as shown below, it fails miserably, producing a huge error message: cout << accumulate(v.begin(), v.end(), string(), if_then_else_return(_1 == string(), _1, _1 + bind(constructor<string>(), var(i)++, ' ') + _2)) << endl; Now for the questions: What is the proper way to use if_then_else_return? Why do I need to define a separate arithmetic action for strings, even though they are a common library type? Is there a way to say the most intuitive: cout << accumulate(v.begin()+1, v.end(), v.front(), _1 + ", " + _2) << endl; Is there way to do more intuitive: cout << accumulate(v.begin()+1, v.end(), v.front(), _1 + string(i++, ' ') + _2) << endl; Thank you for your help. Robert Ioffe [Non-text portions of this message have been removed]

1 0

prim_minimum_spanning_tree crashes
by Stephan Höfer 06 Aug '03

06 Aug '03

Hello, I try to use prim_minimum_spanning_tree to find rings in subgraphs doing the following: typedef subgraph< adjacency_list<vecS, vecS, undirectedS, VertexProperty, EdgeProperty> > Graph; /... building Graph g with some vertices and edges for(int i=0; i<7; ++i) { add_vertex(g); LAMath::Vec3_f tVec((float)i, (float)(6-i), .0); put(vertex_point_t(), g, i, tVec); } property_map<Graph, edge_weight_t>::type weightmap = get(edge_weight, g); Edge e; bool inserted; tie(e, inserted) = add_edge(0, 1, g); weightmap[e] = 1; tie(e, inserted) = add_edge(1, 2, g); weightmap[e] = 1; tie(e, inserted) = add_edge(2, 3, g); .../ Graph tSubGraph = g.create_subgraph(); add_vertex(1, tSubGraph); add_vertex(3, tSubGraph); add_vertex(4, tSubGraph); add_vertex(6, tSubGraph); std::vector < graph_traits < Graph >::vertex_descriptor > p(num_vertices(tSubGraph)); prim_minimum_spanning_tree(tSubGraph, &p[0]); After having created the graph g, I call the create_subgraph member function to get an induced subgraph of it. Trying to use the prim_minimum_spanning_tree function it crashes in dijkstra_dispatch1 by trying to delete the local variable std::vector<D> distance_map(n). Is there something missing in my code which leads to that crash? It works for the root graph but not for any new created subgraph. Help would be appreciated. Thanks in advance Stephan

4 8

boost and vxWorks?
by Alexander Nasonov 06 Aug '03

06 Aug '03

Does anybody use boost libraries under Vxworks? I'm especially interested in pool and threads libraries. Thankin advance -- Alexander Nasonov Remove minus and all between minus and at from my e-mail for timely response

2 1