- Boost-users - lists.boost.org

Tuples and stringstream
by CodeLogic 11 Apr '05

11 Apr '05

Hi, I'm trying to use stringstreams in a tuple like fstream fin(filename.c_str(), ios_base::in); tuple<stringstream, stringstream, stringstream, stringstream> test; fin>>test; I also just tried declaring a variable of type: tuple<stringstream> test; However, both, in MS VC++ 7.0 give me the following error: c:\.....\include\boost-1_32\boost\tuple\detail\tuple_basic_no_partial_spec.hpp(190): error C2558: class 'std::basic_stringstream<_Elem,_Traits,_Alloc>' : no copy constructor available or copy constructor is declared 'explicit' with [ _Elem=char, _Traits=std::char_traits<char>, _Alloc=std::allocator<char> ] I'm basically looking for a solution of being able to stream tuples with strings in the them in and out of a file. Any suggestions will be appreciated. Thanks !

2 4

[BGL] Problem with counting paths/custom dfs_visitor
by Pete Donnell 11 Apr '05

11 Apr '05

Hi, I hope you will forgive me for asking a rather simple question. I have been having a lot of trouble writing a visitor for use in depth first search. The algorithm I'm attempting to write is supposed to count the number of different paths between two vertices on a directed graph. The graph may or may not contain cycles, and of course walks that contain cycles are not counted as paths. The way I've approached the problem is to induce a subgraph on the source and sink vertices, by copying the graph and then removing any vertices that do not lie on a path between the source and sink. My plan then is to do a depth first search on the induced subgraph, using a custom visitor that counts the number of forward or cross edges, which I think will be a count of the number of paths between the source and sink. To be honest, I haven't thought too hard about whether this is algorithmically correct, as the main problem is with the visitor. If I figure out to write custom visitors then it shouldn't be too hard to write other algorithms. As things stand, the function to induce a subgraph is working fine (although I am not using the Boost subgraph, mainly because it isn't documented in the book and so I didn't become aware of its existence until after I'd written the code). Incidentally, this function is useful in itself, which is why I have used it. I'm aware that it's not a very efficient way of solving the problem. The code I've written for the visitor to call on the path counting algorithm currently goes like this (I have slightly modified it a lot of times to try and get it to work): template<typename OutputIterator> class EnumeratorVisitor : public boost::dfs_visitor<OutputIterator> { private : long unsigned int pathCount; public : EnumeratorVisitor() : boost::default_dfs_visitor(), pathCount(1) {} EnumeratorVisitor(OutputIterator out) : boost::dfs_visitor<OutputIterator> (out), pathCount (1) {} long unsigned int getPathCount() const { return pathCount; } template <typename Graph, typename Edge> void forward_or_cross_edge(const Edge& e, const Graph& g) { ++pathCount; } }; The code where I use it looks like this: template<typename Graph> long unsigned int enumeratePaths(const Graph& graph, const typename boost::graph_traits<Graph>::vertex_descriptor& startNode,[BGL] problem with const typename boost::graph_traits<Graph>::vertex_descriptor& endNode) { typedef typename boost::graph_traits<Graph>::edge_descriptor Edge; std::vector<Edge> forwardOrCrossEdges; // vector used to store the forward or cross edges Graph* tempGraph = generateSubGraph(graph, startNode, endNode); // this function induces a subgraph as previously mentioned typedef boost::color_traits<boost::default_color_type> Colour; std::vector<boost::default_color_type> colourMap(boost::num_vertices(*tempGraph), Colour::white()); EnumeratorVisitor<typename std::back_insert_iterator<typename std::vector<Edge> > > edgeInserter(std::back_inserter(forwardOrCrossEdges)); depth_first_visit(graph, startNode, edgeInserter, boost::make_iterator_property_map(colourMap.begin(), boost::get(boost::vertex_index, *tempGraph)) ); return edgeInserter.getPathCount(); } I'm sure there are obvious errors in there to those who know what they're doing, but I've been looking at this problem for over a month now without success, so it's got to that stage where I don't really know what's going on anymore. I've hunted through the book, online code and so on, but I can't figure it out. It gives about three pages of template errors at the moment! There have been other minor variants, but all of them have given errors of some kind. If anyone can point out what I'm doing wrong, I'd greatly appreciate it. If more information is needed, I'll happily provide it. By the way, I'm sure that this problem of path counting has been approached before, it anyone has any insights to offer, I'd also appreciate that! Many thanks for your time, Pete

1 1

Bad adjacency_iterator with -O2, -m64 and gcc 3.2.2 on an opteron
by Michal Simecek 11 Apr '05

11 Apr '05

Hi, compiling and linking the following program on a opteron with the following command and the operating system red hat 3 (current enterprise server version) gcc-3.2.2 -c -I. -I../source -I./boost_1_32_0 -O2 -m64 boosttest.cc -o boosttest.o gcc-3.2.2 -m64 -o boosttest boosttest.o -lstdc++ will cause, that the program runs into an endles loop. I figured out, that the pointer in VIter which is compared when comparing VIter and VIterEnd is damaged. With a lot of differen gcc versions and also in 32 bit mode and with lower optimization it works fine. The problem is, I have to run it also compiled with gcc 3.2.2. Is it an optimization bug in the compiler or any mistake I made or a problem in the boost? I was looking for a hint on the net but without success. Can anybody help? Thanks ------------------------------------------------------------------------------- int main (int argc, char **argv) { typedef boost::adjacency_list < boost::listS, boost::listS, boost::undirectedS > my_graph2_t; my_graph2_t MyGraph2; boost::graph_traits<my_graph2_t>::vertex_descriptor MyVDescriptor; boost::graph_traits<my_graph2_t>::vertex_descriptor MyNextVDescriptor; MyVDescriptor = boost::add_vertex(MyGraph2); MyNextVDescriptor = boost::add_vertex(MyGraph2); boost::add_edge(MyVDescriptor, MyNextVDescriptor, MyGraph2); boost::graph_traits<my_graph2_t>::adjacency_iterator VIter; boost::graph_traits<my_graph2_t>::adjacency_iterator VIterEnd; tie(VIter, VIterEnd) = boost::adjacent_vertices(MyVDescriptor, MyGraph2); while ( VIter != VIterEnd ) { std::cout << "loop ..." << std::endl; ++VIter; } return 0; } --------------------------------------------------------------------------------

2 2

Re: You provide gun, I provide foot.
by C. Michailidis 10 Apr '05

10 Apr '05

> > Answer.. Don't take the examples at their face value Likewise. > > you can do this: > > > > class counter > > { > > public: > > counter() : guard(this->mutex) { > > this->count = 0; > > this->guard.unlock(); > > } > > void lock() { > > guard.lock(); > > } > > int increment() { > > assert(this->guard.locked()); > > return ++count; > > } > > void unlock { > > assert(!this->guard.locked()); > > this->guard.unlock(); > > } > > // ..Nothing to do here.. > > // ~counter() { > > // } > > > > private: > > boost::mutex mutex; > > boost::mutex::scoped_lock guard; > > int count; > > }; > > This is not equivalent to the above scenario. You can't have two threads > calling lock() at the same time, because a scoped_lock object (as currently > specified) is not thread safe (whereas a mutex is, by its very nature.) > Ugh, it would seem you are right Peter! Again, I quote the docs: "sharing a lock object between threads results in undefined behavior". Phooey! > > > class counter > > { > > public: > > counter() : count(0) { > > } > > void lock() { > > // nothing, methods act as self monitors > > } > > int increment() { > > boost::mutex::scoped_lock guard(this->mutex); > > return ++count; > > } > > void unlock { > > // nothing, methods act as self monitors > > } > > // ..Nothing to do here.. > > // ~counter() { > > // } > > > > private: > > boost::mutex mutex; > > int count; > > }; > > This is also not quite equivalent to the original. In it, a lock() op1() > op2() unlock() is a transaction; no other thread can inject its own op3 > between op1 and op2. Exactly - this is why I said "likewise" earlier, d-oh! > The closest you can get to the original, while somewhat respecting the > requirement for it to be callable from outside C++, is: > > scoped_lock * lock() > { > return new scoped_lock( mutex ); > } > > void unlock( scoped_lock * lock ) > { > delete lock; > } Although this respects the requirement that the code be callable from outside C++, it completely ignores the requirement that the COM interface (i.e. method prototype) should not change :-( Anyway, the last suggestion (changing the COM interface and returning a pointer to the scoped_lock) is probably the best I can do for now - I have indeed already considered this. The practitioner in me says it should be fine for my purposes. The theorist in me says I should keep moaning and groaning until someone truly feels my pain. soap_box->lock(); I've been making a stink largely because (IMHO) the decision to NOT expose functionality as fundamental as mutex.lock() and mutex.unlock() seems ridiculous! The boost library and accompanying docs should absolutely encourage the use of scoped_locks - I agree 100%. But it certainly shouldn't *disallow* me from using mutexes in the traditional way - to me this is a shortcoming. Even if there is some kind of overly complicated way (perhaps by deriving from mutex and calling some protected lock/unlock methods) I still think lock/unlock functionality should be exposed by mutex objects. Did you ever hear of the expression "you've polished the brass right off of the hinges"? This honestly seems like arrogance on the boost development team's part - why do they know how to use mutexes so much better than the rest of us? What am I, chopped liver over here? It is belittling! I'm not a child, whaaaaa!!!! ;-) There is little sense in going to such extreme lengths to protect developers from themselves - after all they are the ones writing the programs! Believe me, I would love to live to see the day that I simply install the Boost libraries and BINGO by some stroke of genius the program I intended to write is already done. I have a feeling it will be some time before this happens, until then my gun should shoot bullets wherever I point it, and I should be allowed to point it wherever I want (not where the boost developers THINK it should be pointed, but where I KNOW it should be pointed). As they say, "guns don't kill people, people kill people", lol. soap_box->unlock(); Just my 2 cents, C. Michailidis

4 4

Re: You provide gun, I provide foot.
by C. Michailidis 10 Apr '05

10 Apr '05

> Whining won't take you far on this list, sorry. Call it arrogance, call it > indifference, whatever -- don't forget that Boost authors have volunteered > enormous amounts of their own time. Seriously, you have the boost source, if > you don't like the implementation of something in threads go change it and > propose it. Or just use it for yourself. That will get you much further than > soapbox grandstanding any day... > > BTW, I think the question of what to expose was discussed in the library > review, so you might search back thru the mailing lists. Also you should know > that the original threads author has disappeared and the new maintainer is > pretty busy -- so I'd say threads is somewhat undersupported at the moment... > > Jeff Yeah, I've got to agree - you make good points here! However, you have to admit, ranting and raving will (occasionally) make you FEEL better, LOL. In any case, the boost libs still rock! Maybe I'll dig a little deeper and make a proposal; after April 15th of course! Cheers, Dino P.S. Thanks to everyone who put up w/ my moaning and groaning. I just wanted to make sure I wasn't overlooking something dumb, which I will do every now and again ;-)

1 0

Re: You provide gun, I provide foot.
by C. Michailidis 10 Apr '05

10 Apr '05

> do have models. What prevents you from using say, the > boost::mutex::scoped_lock model? (Which has the lock/unlock calls.) Perhaps my initial mailing was too general. I'll try to state my problem more specifically. The following url, http://www.boost.org/libs/thread/example/mutex.cpp , gives an example of how mutexes and scoped_locks are typically used. The counter class shows how an atomic counter is implemented. For completeness I'll copy and paste part of the example into my email, here's the class: class counter { public: counter() : count(0) { } int increment() { boost::mutex::scoped_lock scoped_lock(mutex); return ++count; } private: boost::mutex mutex; int count; }; As I understand it, a thread calling the 'increment' method will be blocked until any other thread that has already called 'increment' has returned from the method call (more or less). When the scoped_lock object is instantiated its constructor locks the mutex until the scoped_lock goes out of scope (at which time the destructor of scoped_lock unlocks the mutex enabling other threads to reach the body of increment). This is all fine and dandy. My problem is that I have an existing COM component which is more similar to: // please excuse the MSFT parlance class counter { public: counter() : count(0) { mutex=CreateMutex(NULL, false, NULL); } void lock() { WaitForSingleObject(mutex, INFINITE); } int increment() { return ++count; } void unlock { ReleaseMutex(mutex); } ~counter() { CloseHandle(mutex); } private: unsigned long mutex; int count; }; When using the COM component the caller is required to call 'lock' before calling any other methods (in this case 'increment'), when you are finished using the component you must call 'unlock'. So... here's my problem, how can I use Boost.Threads to create the same kind of functionality without changing my COM interfaces? Remember that this is just for illustration so my sample code may need changes before compiling ;-) In any case, the actual component is more sophisticated than the simple example I gave, and could actually be called from a language other than C++ (e.g. Visual Basic). AFAIK boost mutexes must be locked using scoped_locks and it would seem that they aren't appropriate for my situation. I can certainly appreciate the benefits of scoped_locks (esp. in a homogeneous environment where all code is written in C++). It just seems silly to force the user to use ONLY scoped_locks when in some situations the additional requirements they have make them unusable. Sure, most of the time the mutex should be unlocked at the end of the block... however when the critical section is composed of code written in more than one language or in a more dynamic algorithm it may easier to unlock it elsewhere, no? Perhaps this is an issue more appropriate for the boost-devel list... but, if there really isn't a way to simply lock/unlock a mutex I think it is awefully rude and arrogant of the developers considering the fact that more straight-forward multi-threaded IPC mechanisms have been around for quite some time and programmers have been able to use them effectively without much trouble. It's my opinion that a great toolkit not only provides robust and sophisticated facilities but *ALSO* gives the developer enough 'rope to hang themselves'. After all, isn't the freedom to shoot yourself in the foot one of the major advantages of writing code in C/C++? I use the Boost library not only to add functionality to my software but to act as a common API in cross-platform environments. Nearly every modern API supports some kind of fundamental lock/unlock mechanism for mutexes, why shouldn't boost? And yes, to answer one of the replys... I am not being forced to use boost, in this case it seems as though I can't! It would be nice however, since I'm using other Boost modules in the project. In summary, since "Lock concepts are the only way to lock and unlock a mutex object: lock and unlock functions are not exposed by any Boost.Threads mutex objects", it seems as though I'm left out in the cold! Any suggestions? Any boost developers reading this? Anyone feel my pain? Am I just being blindly stubborn? LOL, Dino

3 3

Heterogenous properties
by Martin Jynge 10 Apr '05

10 Apr '05

Hi. I'm trying to develop a (bidirectional) graph whose edges contain a property Value, which is a baseclass with different inherited classes dependent upon the type of vertex they are connected to. I attach a pointer to a base class Node to each vertex - the node containing only a virtual method called from a designated parsing algorithm. This method needs to access the Value leafclasses of the vertex's outedges. For example, a particular vertex receives a double from one outedge and a struct xy {double x,y; } from another, does some operation and stores, say, a complex number in a relevant inedge. In essense it's kind of like genetic programming with multiple parents and multiple value types. I would really like to avoid using dynamic_cast in dispatching edge Value descendants. Any thoughts on this?

1 0

BOOST_CHECK_EQUAL compiler warning
by shopper11＠cfl.rr.com 09 Apr '05

09 Apr '05

I get a compiler warning when using BOOST_CHECK_EQUAL in conjunction with BOOST_AUTO_UNIT_TEST. The code and warning are below. Anyone know what I'm doing to cause the warning? Environment: Windows XP with MS VS 2003 #define BOOST_AUTO_TEST_MAIN #include <boost/test/auto_unit_test.hpp> class MyTest { public: MyTest( int newVal = 3 ) : a (newVal) {} int GetA() {return(a);} protected: int a; }; MyTest test1; // _____________________________________ BOOST_AUTO_UNIT_TEST( TestCase1 ) { int b = test1.GetA(); BOOST_CHECK_EQUAL( b, 1 ); } The warning is: c:\Boost\include\boost-1_32\boost\test\detail\wrap_stringstream.hpp(69) : warning C4267: 'argument' : conversion from 'size_t' to 'unsigned int', possible loss of data c:\Boost\include\boost-1_32\boost\test\test_tools.hpp(435) : see reference to function template instantiation 'boost::basic_wrap_stringstream<CharT> &boost::operator <<<char,size_t>(boost::basic_wrap_stringstream<CharT> &,const T &)' being compiled with [ CharT=char, T=size_t ] m:\jclaybau_InWork_view\CHAMELEON\SoftwareDevelopmentFolder\Source\AM\Test\AMTest\AMTest.cpp(20) : see reference to function template instantiation 'bool boost::test_tools::tt_detail::equal_and_continue_impl<int,int>(const Left &,const Right &,boost::wrap_stringstream &,boost::test_tools::const_string,size_t,boost::unit_test::log_level,size_t)' being compiled with [ Left=int, Right=int ]

2 1

Followup to KCC compile error
by Stephen Gross 09 Apr '05

09 Apr '05

To review: > When including "boost/iterator/indirect_iterator.hpp", I get: > > "/usr/local/sagedev/apis/include/boost_1_32_0/boost/detail/is_incrementable.hpp", > line 56: error: > > expression must have arithmetic, enum, or pointer type > > BOOST_STATIC_CONSTANT( > ^ Some new thoughts upon investigation: It is not in the BOOST_STATIC_CAST. Doing manual compiler preprocessing has resulted in the following code, which is what the compiler is complaining about: template <class T> struct impl { static typename remove_cv<T>::type& x; static const bool value = sizeof(is_incrementable_::check(++x,0)) == 1; }; The problem is the ++x, not the static cast. The problem seems to be a question of that x variable. If I replace the typename of x with, say int&, it works with one additional modification (see below), and if I replace it with a non-incrementable object, say 'foo', that also works. My theory was it might be in the remove_cv, but replacing the type with T& does not work. We still get the 'must be of XXX' type message despite T being known to be a type. So, it would seem to be something about the templates. I'm stuck at this point. What I had to do to make known types work: With known types (int, foo), the compile only works when I add an additional set of parentheses around the ++x,0 in the check. I assume this is an order of evaluations thing, to make certain the comma operator is processed before the check function, which since all versions of check are single argument, I assume is the intent here. --Steve (sgross(a)darwin.cwru.edu)

2 1

potential bugs in BGL function "brandes_betweenness_centrality()"
by Qiaofeng Yang 09 Apr '05

09 Apr '05

Hi, I use g++ (GCC) 3.2.2 20030222 (Red Hat Linux 3.2.2-5). What I did is as follows: ===================================================================== 1. Use "brandes_betweenness_centrality()" to calculate the vertex betweenness centrality for all the vertices in the graph. 2. Find the vertex with the highest betweenness value. 3. Delete the vertex from the graph. 4. Repeat step 1 until no vertex left in the graph. ===================================================================== The code is sketched as follows: ======================================================================== typedef property<vertex_name_t, string> VertexName; typedef property<vertex_index_t, int, VertexName> VertexProperty; typedef property<edge_index_t, int> EdgeProperty; typedef adjacency_list<listS, listS, undirectedS, VertexProperty, EdgeProperty> Graph; Graph G; //load graph ... 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(...); } ======================================================================== The above code is run on a graph with 8 vertices and 9 edges. Below is the edge list representation of my graph: node0 node4 node0 node1 node0 node2 node0 node3 node1 node2 node2 node3 node4 node5 node4 node6 node4 node7 The problem is as follows: 1. Extremely slow. The first iteration is fast, but you have to wait for a long time before the results of the second iteration to print out. 2. After several iterations, the vertex name printed out is wrong. (In my graph, at the third iteration, the vertex name is printed out wrong) 3. When there are two vertices and no edges in the graph, the program stops printing out anything, basically freezes there. >From my previous experience with BGL, the subgraph function has similar problem. When I create a subgraph from a graph, everything is inherited correctly, including vertex names, edges, etc. When I try to create a subgraph from the subgraph from the previous step, I get segmentation fault when trying to print out vertex names. I think there may be some bugs in the source files dealing with the graph structure modification. Best regards, Qiaofeng

2 1