- Boost-users - lists.boost.org

RE: [Boost-users] A Simple Problem: File Transfer...
by Sliwa, Przemyslaw (London) 22 Apr '05

22 Apr '05

Are there any plans to include the networking libraries into boost? I have heard about the Boost.Socket but I guess this project has been abandoned long time ago. Is it correct? Pshemek -----Original Message----- From: boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org] On Behalf Of Ben Hutchings Sent: 18 April 2005 17:36 To: boost-users(a)lists.boost.org Subject: Re: [Boost-users] A Simple Problem: File Transfer... Hossein Haeri wrote: > Dear all, > > If you were assigned to write a file transfer utitlity > "in C++", what would your suggestions be? More > specific, I've got two IP's (perhaps plus their > ports), I've got to send one file from one to the > other. Which parts of the Boost would you use, if any? Boost probably isn't going to help you much with file transfer, as it doesn't (yet) include any networking libraries. There are various Boost libraries that might be useful, though, such as smart_ptr and program_options. Ben. _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users -------------------------------------------------------- If you are not an intended recipient of this e-mail, please notify the sender, delete it and do not read, act upon, print, disclose, copy, retain or redistribute it. Click here for important additional terms relating to this e-mail. http://www.ml.com/email_terms/ --------------------------------------------------------

9 12

[multi_index] FYI: Experiences with the Boost.MulitIndex container
by Jochen Hammann 22 Apr '05

22 Apr '05

Hi, I am using the Boost MultiIndex container in conjunction with the Microsoft Visual C++ 7.1 compiler. I encountered an runtime exception during some tests of my application. I made the mistake, that I tried to construct an index as a separate object instead of using a reference to the respective index. The MultiIndex documentation highlights this issue: "Note that get returns a reference to the index, and not an index object. Indices cannot be constructed as separate objects from the container they belong to, so the following // Wrong: we forgot the & after employee_set::nth_index<1>::type const employee_set::nth_index<1>::type name_index=es.get<1>(); does not compile, since it is trying to construct the index object name_index. This is a common source of errors in user code." I could reproduce the exception by using the MultiIndex example basic.cpp from the Boost 1.32.0 release. Unfortunately, Microsoft Visual C++ 7.1 compiles the above mentioned line without an error or warning. During the execution of the example an exception will be thrown while processing the following line: std::copy(i.begin(),i.end(),std::ostream_iterator<value_type>(std::cout)); within the function template print_out_by(). Regards, Jochen.

4 13

boost::serialization: how to BOOST_CLASS_VERSION a template class?
by Bo Peng 22 Apr '05

22 Apr '05

Dear list, I use boost::serialization to serialize my classes. They are template classes that are instantiated and wrapped by SWIG. Everything works fine until I decide to use BOOST_CLASS_VERSION to set version for the class. The first parameter of BOOST_CLASS_VERSION(class, version) is a class but my template will only be instantiated by SWIG. The syntax does not allow me to do something like template<class T> BOOST_CLASS_VERSION(class<T>, version) or BOOST_CLASS_VERSION( template<class T>class <T>, version). What should I do? Many thanks in advance. Bo

2 2

Question about memory management and Serialization
by kauai 22 Apr '05

22 Apr '05

Hi folks, When I unserialize an object that has pointers in it (composite recursive structure), does boost unserialize take care of allocating them on the heap? I assume it does, however... I run a memory profiling tool call 'valgrind' and I on purpose decided not to cleanup my newly created serialized object. What happened after running through valgrind surprised me, it reported no memory leaks. Whats the magic here, smart pointers? Thanks for any help, Kauai __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

2 1

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