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

Re: potential bugs in BGL function "brandes_betweenness_centrality()"
by qyang＠cs.ucr.edu 22 Apr '05

22 Apr '05

Hi Doug, Thanks for the hint! I manually re-index all the vertices after the vertex removal(a for() loop) and put vertex_centrality(num_vertices(G)) outside the while() loop. The code works fine now :). I have another question concerning vertex and edge iterator. The problem is now that I want to remove the nth edge in the graph. I use the following code: ================================================== //typedef stuff ... EdgeDescriptor Edge; Edge = *(edges(G).first + n); //n is an int remove_edge(Edge, G); ================================================== I got compilation errors at line "Edge = *(edges(G).first + n);". I have to use the following code in order to compile the code and to run it correctly: ================================================== //typedef stuff ... EdgeDescriptor Edge; EdgeIterator edge_iter; edge_iter = edges(G).first; for(int i = 0; i < n; i++) { edge_iter++;} Edge = *edge_iter; remove_edge(Edge, G); ================================================== Basically, you have to use a for() loop to increment the edge_iterator varialbe. Is it possible just to use one addition(the first code) instead of n additions(the second code) to increment the iterator? Best regards, Qiaofeng >> while(num_vertices(G)>0) >> { >> int V = num_vertices(G); >> vector<double> vertex_centrality(V); >> >> brandes_betweenness_centrality(G, >> make_iterator_property_map(vertex_centrality.begin(), get(vertex_index, >> G), double())); >> >> //print the vertex betweenness centrality for all the vertices >> ... >> >> //find the index of the vertex with the highest betweenness centrality >> ... >> >> clear_vertex(...); >> remove_vertex(...); >> } > > After the call to remove_vertex, the vertex indices are no longer > valid, e.g., > > 1) The first time through the loop, V = 9. The vertices have indices 0, > 1, 2, 3, 4, 5, 6, 7, 8 > 2) We delete one vertex, so V = 8. Say it's the vertex with index 3. > The vertices have indices 0, 1, 2, 4, 5, 6, 7, 8 > > During the execution of brandes_betweenness_centrality in the second > iteration, we'll try to access vertex_centrality[8], which is an error. > vertex_centrality has been reallocated to only allow 8 elements, not 9. > > The easiest way to fix this is to remove the vertex_centrality vector > from the loop, then add: > > vector<double> vertex_centrality(num_vertices(G)); > > *before* the loop. This will speed up the computation because the > vector doesn't get reallocated each time. More importantly, it will > stay at the maximum number of vertices, so even though we delete vertex > index "3" there is still a vertex index "8". The vector will have some > "holes" in it where values aren't used after the first iteration, but > that's okay. > > Vertex and edge indices are a weak point in the BGL. It seems like they > should be automatically updated, but that can't be done efficiently. We > have a partial solution (a graph type that is reasonably good > all-around and does automatic (re-)indexing), but it's not ready for > release yet. > > Doug

2 1

Library locations for Thread Lib example
by Rush Manbert 21 Apr '05

21 Apr '05

Hi, I build the Boost libraries and installed them, and when I did I specified --prefix, --exec-prefix/,/ and / /--builddir so that all of my build products would be placed in a separate directory tree. I have been looking at the thread library. I could run the test as described, using bjam -sTOOLS=darwin test in the proper directory. It built and ran with no problems. (This is under Mac OS 10.3.8) Flushed with success, I built the thread example programs using this command: bjam -sTOOLS=darwin This did build the programs, placing them under $BOOST_ROOT/bin/boost/libs/thread/example/. For instance, the "condition" program is located in: $BOOST_ROOT/bin/boost/libs/thread/example/condition/darwin/debug/ All okay. Here is my problem. When I try to run these programs, they all fail. A representative error is: dyld: ./condition/darwin/debug/condition can't open library: ../../../bin/boost/libs/thread/build/libboost_thread.dylib/darwin/debug/shared-linkable-true/libboost_thread-d-1_32.dylib (No such file or directory, errno = 2) It appears to me that the link knows the relative path to the default library build location, and specified that as the path to the dynamic library. The problem is that I didn't build my libraries there. I tried setting BOOST_ROOT to the equivalent top level location in my build products tree, but that didn't change anything. I tried to force a static link (hoping to force a different link error that might give me another hint), but that blew up bjam in a different way that I don't want to get into here. (Should I be able to statically link these programs?) Is there some way to change the library path used for linking these programs? Thanks, Rush

1 0

scoped_array and legacy APIs
by Sean DeNigris 21 Apr '05

21 Apr '05

Hey-hey, Is it proper to to use static_cast like this: //Library.h - legacy library typedef ??? Type; //library-defined, we don't know the actual type typedef Type* PType; //library-defined pointer type //sets bufSize to # bytes necessary if buf not big enough LibraryFunction(PType buf, int* bufSize); //MyCode.cpp int bytesNeeded = 0; LibraryFunction(0, &bytesNeeded); scoped_array<Type>(static_cast<PType>(new BYTE[bytesNeeded])); The legacy library tells us storage requirements in BYTEs, but I want to deal with 'Type's, not BYTEs. Is there a conceivable reason why this could not be safe, even if it happens to be with the current implementation? Thanks. - Sean

4 3

Patch for memory leak in boost thread library (tss)
by Bryan Tower 21 Apr '05

21 Apr '05

The cleanup_slots method called when threads exit appears to be leaking a tss_slots object. This causes a leak per thread that uses the thread specific storage. The following patch prevents this per thread memory leak. bryan diff -c -r1.1 tss.cpp *** boost_1_32_0/libs/thread/src/tss.cpp 8 Sep 2004 15:58:40 -0000 1.1 --- boost_1_32_0/libs/thread/src/tss.cpp 21 Apr 2005 00:15:15 -0000 *************** *** 56,61 **** --- 56,62 ---- (*tss_data->cleanup_handlers[i])((*slots)[i]); (*slots)[i] = 0; } + delete slots; } void init_tss_data()

1 0

[Python] Embedding and Extending problem
by Keith MacDonald 21 Apr '05

21 Apr '05

I've tried to merge the Boost.Python tutorials for embedding and extending (see below), to end up with an executable that does both. The script fails to import "hello", so BOOST_PYTHON_MODULE is presumably not the correct way to do it. What is? Thanks, Keith MacDonald ====== The C++ code ====== #include <boost/python.hpp> using namespace boost::python; char const* greet() { return "hello, embedded world"; } BOOST_PYTHON_MODULE(hello) { def("greet", greet); } int main() { Py_Initialize(); object main_module(( handle<>(borrowed(PyImport_AddModule("__main__"))))); object main_namespace = main_module.attr("__dict__"); try { handle<> ignored((PyRun_String( "import hello\n" "print hello.greet()\n" , Py_file_input , main_namespace.ptr() , main_namespace.ptr()) )); } catch(error_already_set) { PyErr_Print(); } Py_Finalize(); return 0; } ====== The Jamfile ====== subproject libs/python/example/tutorial/embedded ; # location of the program # bring in the rules for python SEARCH on python.jam = $(BOOST_BUILD_PATH) ; include python.jam ; exe hello # name of the executable : #sources hello.cpp : # requirements <find-library>boost_python <library-path>c:\boost\libs\python $(PYTHON_PROPERTIES) <library-path>$(PYTHON_LIB_PATH) <find-library>$(PYTHON_EMBEDDED_LIBRARY) ;

2 4

boost::regexp_iterator illogical behaviour?
by Sandor 21 Apr '05

21 Apr '05

Hi, I am using the boost::regexp_iterator to list every match of a user configured regexp in the contents of a file. The file is read up into a single string and passed into the constructor of regexp_iterator. Take the example regexp of "a|\Ab" which means that every 'a' letter will be a match, plus the 'b' letter should be a match if it is the first one of the file. But the algorithm does not work like this! The letter 'b' is matched every time it follows an 'a'. Because '\A' does not mean the beginning of the original client-code-supplied buffer! After the first match, the '\A' only means the end of the last match. I can imagine situations where a metacharater with such meaning is needed, but I would need a different behaviour. Diving into the boost source code: template <snip> class regex_iterator_implementation { <snip> bool next() { if(what.prefix().first != what[0].second) flags |= match_prev_avail; BidirectionalIterator next_start = what[0].second; match_flag_type f(flags); if(!what.length()) f |= regex_constants::match_not_initial_null; bool result = regex_search(next_start, end, what, *pre, f); if(result) what.set_base(base); return result; } <snip> } What I would think logical is (note the two new lines): template <snip> class regex_iterator_implementation { <snip> bool next() { if(what.prefix().first != what[0].second) flags |= match_prev_avail; BidirectionalIterator next_start = what[0].second; match_flag_type f(flags); if(!what.length()) f |= regex_constants::match_not_initial_null; if(base != next_start) f |= regex_constants::match_not_bob; bool result = regex_search(next_start, end, what, *pre, f); if(result) what.set_base(base); return result; } <snip> } Or at least provide a new flag to enable this behaviour. It is quite tedious to reimplement regex_iterator in client code just to add this small feature. What do you think? - Sandor

2 1

[Simple example] - Compile error in io_fwd.hpp (VC6)
by Romain Prieto 21 Apr '05

21 Apr '05

Hello, I'm new to Boost, and I've just tried to build the serialization library. I've downloaded Boost, and run the following command: >> C:\....\Boost\bjam "-sTOOLS=msvc" install (I got a few warnings about failed builds) Now, when I compile a simple example from the website, i get the following error (many times): c:\...\boost\io_fwd.hpp(28) : error C2239: unexpected token '>' following declaration of 'Ch' c:\...\boost\io_fwd.hpp(28) : error C2059: syntax error : '>' c:\...\boost\io_fwd.hpp(28) : error C2371: 'Ch' : redefinition; different basic types c:\...\boost\io_fwd.hpp(28) : see declaration of 'Ch' Does it have something to do with VC6 ? (I know it's not the best compiler, but I have no choice) I've already installed STLport to have a better template-support... Thank you Romain

1 0

sparse matrix iterator problem
by Golan Green 21 Apr '05

21 Apr '05

Hi, I am trying to use sparse matrix iterators, but it seems that a code that works in VC++ 6.0 does not compile under C++ in .NET I tried a simplified version of the code and it still does not work. boost::numeric::ublas::sparse_matrix<int> matrix; boost::numeric::ublas::sparse_matrix<int>::iterator1 it1 = matrix.begin1(); boost::numeric::ublas::sparse_matrix<int>::iterator2 it2 = it1.begin(); The error message that I get is : error C2039: 'begin' : is not a member of 'boost::numeric::ublas::sparse_matrix<T>::iterator1' with [ T=int ] off course begin() is one of the iterator1 member functions. Does anyone have an idea of what is missing in the project settings? Regards, Golan. --

1 0

RE: [Boost-users] Re: why so many .so and .a files for boost?
by Tom Tan (SH/RDC) 21 Apr '05

21 Apr '05

Yes, to enable multi-threaded code. But how? What's going on inside while compiling with '-pthread'? Or what's the difference between them at the compilation level? Or Is there anything happen at optimization level? -----Original Message----- From: boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org] Sent: Thursday, April 21, 2005 03:59 To: boost-users(a)lists.boost.org Subject: Re: [Boost-users] Re: why so many .so and .a files for boost? On 4/20/05, Tom Tan (SH/RDC) <tom.tan(a)ericsson.com> wrote: > got it, then why do i have to turn on the '-pthread' option of g++ when using boost_thread library? Because you need that option to enable multi-threaded code. The Boost Threads library isn't much use without it :-) -- Caleb Epstein caleb dot epstein at gmail dot com _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

2 1

shared_ptr/weak_ptr compile error
by Rob Benton 21 Apr '05

21 Apr '05

I'd appreciate it if someone could tell me how to debug the error I'm getting. I've got a class with a member function like this: bool Instruction::button_press(GdkEventButton *, boost::weak_ptr<Instruction>, boost::weak_ptr<Instruction>, boost::weak_ptr<Instruction>, boost::weak_ptr<Instruction>); =========================================================================== Instruction.C: In member function `bool Instruction::button_press(GdkEventButton*, boost::weak_ptr<Instruction>, boost::weak_ptr<Instruction>, boost::weak_ptr<Instruction>, boost::weak_ptr<Instruction>)': Instruction.C:138: error: `top' has incomplete type /usr/include/boost/shared_ptr.hpp:43: error: declaration of `struct boost::weak_ptr<Instruction>' Instruction.C:138: error: `left' has incomplete type /usr/include/boost/shared_ptr.hpp:43: error: declaration of `struct boost::weak_ptr<Instruction>' Instruction.C:138: error: `right' has incomplete type /usr/include/boost/shared_ptr.hpp:43: error: declaration of `struct boost::weak_ptr<Instruction>' Instruction.C:138: error: `bottom' has incomplete type /usr/include/boost/shared_ptr.hpp:43: error: declaration of `struct boost::weak_ptr<Instruction>' =========================================================================== Thanks, Rob

2 2