May 2003 - Boost-users - lists.boost.org

BGL leaky?
by Martin Okrslar 06 May '03

06 May '03

Dear Boost users, is the BGL leaky? For example the code added below produces memory leaks (as seen from also added valgrind output). When I comment out the line marked by /* THIS LINE DOES IT */, there are no leaks. I am running BGL 1.29 and gcc 3.2. Any ideas for a work-around? cheers Martin #include <iostream> // for std::cout #include <utility> // for std::pair #include <algorithm> // for std::for_each #include <boost/graph/graph_traits.hpp> #include <boost/graph/adjacency_list.hpp> #include <boost/graph/dijkstra_shortest_paths.hpp> using namespace boost; int main(int,char*[]) { // create a typedef for the Graph type typedef adjacency_list<vecS, vecS, bidirectionalS> Graph; // Make convenient labels for the vertices enum { A, B, C, D, E, N }; const int num_vertices = N; // writing out the edges in the graph typedef std::pair<int, int> Edge; Edge edge_array[] = { Edge(A,B), Edge(A,D), Edge(C,A), Edge(D,C), Edge(C,E), Edge(B,D), Edge(D,E), }; int num_edges = sizeof(edge_array)/sizeof(edge_array[0]); // declare a graph object Graph* g = new Graph(num_vertices); // add the edges to the graph object for (int i = 0; i < num_edges; ++i) add_edge(edge_array[i].first, edge_array[i].second, *g); /* THIS LINE DOES IT */ delete g; return 0; } okrslar@liebrich:~/molgen/domclust/src/test> valgrind --leak-check=yes --show-reachable=yes ./domclust ==23473== valgrind-1.0.4, a memory error detector for x86 GNU/Linux. ==23473== Copyright (C) 2000-2002, and GNU GPL'd, by Julian Seward. ==23473== Estimated CPU clock rate is 2412 MHz ==23473== For more details, rerun with: -v ==23473== ==23473== ==23473== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 0 from 0) ==23473== malloc/free: in use at exit: 384 bytes in 1 blocks. ==23473== malloc/free: 3 allocs, 2 frees, 544 bytes allocated. ==23473== For counts of detected errors, rerun with: -v ==23473== searching for pointers to 1 not-freed blocks. ==23473== checked 6470036 bytes. ==23473== ==23473== definitely lost: 0 bytes in 0 blocks. ==23473== possibly lost: 0 bytes in 0 blocks. ==23473== still reachable: 384 bytes in 1 blocks. ==23473== ==23473== 384 bytes in 1 blocks are still reachable in loss record 1 of 1 ==23473== at 0x4004A3A8: __builtin_new (vg_clientfuncs.c:125) ==23473== by 0x4004A3F6: operator new(unsigned) (vg_clientfuncs.c:138) ==23473== by 0x4035E4F1: ??? (bits/stl_alloc.h:482) ==23473== by 0x4035E40D: ??? (bits/stl_alloc.h:532) ==23473== ==23473== LEAK SUMMARY: ==23473== definitely lost: 0 bytes in 0 blocks. ==23473== possibly lost: 0 bytes in 0 blocks. ==23473== still reachable: 384 bytes in 1 blocks. ==23473== -- ----------------------------------------------------------------------------- Martin Okrslar MPI for Molecular Genetics phone: ++ 49 + 30 / 8413-1166 Computational Molecular Biology Fax: ++ 49 + 30 / 8413-1152 Ihnestrasse 73 email: okrslar(a)molgen.mpg.de D-14195 Berlin URL: http://cmb.molgen.mpg.de -----------------------------------------------------------------------------

2 1

More Mac Compiler Woes - Boost 1.30.0 - Thread and type_traits
by Bobby Thomale 05 May '03

05 May '03

Hey guys, I am having some more problems getting Boost 1.30 to compile on the Mac. I am using CW Pro 8.3, compiling a TARGET_CARBON project. I am having trouble getting the MP implementation of threads to compile. It seems to be having trouble with function pointers where the syntax function<type> is used. I remember at one point it was mentioned that: function<type, type...> was deprecated and we were to use either the more universally supported: functionN<type,type...> or the cleaner: function<void myfuncSig()> instead if our compiler supported it. (Mine doesn't support that last one yet I don't think.) I am guessing that the oldest syntax was removed and no one has bothered to update the Mac carbon thread implementaton to compile again since then? That one seems pretty easy to fix - just update to a supported syntax? (I could do this, update to the functionN<type,type...> syntax since that is what my compiler supports. If someone wants me to make and submit the changes to them for inclusion I'd be happy to, if I knew who to send them to...) Or is there a better/easier workaround? Secondly, the Signals library is not compiling. It's actually not a Signals problem, I think it is a type_traits problem. I am getting a bunch of strange compile errors, in a boost::type_traits file called: is_base_and_derived.hpp It looks like there is a template near the top that is the first thing that fails. There is a define that goes into an alternate implementation for certain broken compilers. Perhaps CodeWarrior needs to be added to the list of compilers that compiles the "broken compiler" version of the stuff in this header? Thanks in advance for your help! -- Bobby --------------------------------------------------------------------- Bobby Thomale Senior Software Developer Inoveon Corporation http://www.inoveon.com/ ---------------------------------------------------------------------

5 6

bug in pthreads implementation of recursive_mutex
by Feldman, Yulik 05 May '03

05 May '03

Hi, It looks that there is a bug in .../boost_1_30_0/libs/thread/src/recursive_mutex.cpp. The bug shows in the pthreads implementation of the recursive_mutex when the mutex is used with the "condition" synchronization primitive. Here is the description of the bug: When the client code calls wait() function on a condition sending as a parameter a lock object associated with a recursive_mutex ("m_Condition.wait(Lock)"), the following functions get called: boost::condition::wait(L& Lock) at condition.hpp:91 boost::condition::do_wait(M& mutex) at condition.hpp:137 boost::detail::thread::lock_ops<boost::recursive_mutex>::unlock(Mutex& m, lock_state& state) at lock.hpp:54 boost::recursive_mutex::do_unlock(cv_state& state) The implementation of the last function on "winthreads" and "mptasks" resets the 'm_count' data member to zero. The "pthreads" implementation at recursive_mutex.cpp:399 doesn't reset the 'm_count' to zero. As a result, when the calling thread enters pthread_cond_wait() and the mutex is unlocked by the pthreads the 'm_count' still remains non-zero. Starting from this point, the things may go bad. Consider, for example, that another thread tries to acquire the mutex. The thread calls: boost::recursive_mutex::scoped_lock Lock(m_Mutex) boost::detail::thread::scoped_lock<boost::recursive_mutex>::scoped_lock<boos t::recursive_mutex>(Mutex& mx, initially_locked = true) at lock.hpp:66 boost::detail::thread::scoped_lock<boost::recursive_mutex>::lock() at lock.hpp:76 boost::detail::thread::lock_ops<boost::recursive_mutex>::lock(Mutex& m) at lock.hpp:34 boost::recursive_mutex::do_lock() at recursive_mutex.cpp:307 The last function locks the mutex again, and, if BOOST_HAS_PTHREAD_MUTEXATTR_SETTYPE is defined, increments the counter to 2. Now, since the count is more than 1, the function unlocks the mutex. As a result, the mutex remains unlocked, after the lock function is called. The same problem exists in recursive_try_mutex and recursive_timed_mutex. I've tried to fix the code as follows, which seemed to help in my particular case: ---------------------------------------------------------------------------- ------------------------------------ *** recursive_mutex.cpp.orig Sun Feb 23 19:38:11 2003 --- recursive_mutex.cpp Sat May 3 17:32:44 2003 *************** *** 412,417 **** --- 412,418 ---- state.pmutex = &m_mutex; state.count = m_count; + m_count = 0; } recursive_try_mutex::recursive_try_mutex() *************** *** 608,613 **** --- 609,615 ---- state.pmutex = &m_mutex; state.count = m_count; + m_count = 0; } recursive_timed_mutex::recursive_timed_mutex() *************** *** 787,792 **** --- 789,795 ---- state.pmutex = &m_mutex; state.count = m_count; + m_count = 0; } #elif defined(BOOST_HAS_MPTASKS) ---------------------------------------------------------------------------- ------------------------------------ Any comments? Is this indeed a bug? Is the fix correct? Thanks, Yulik. [Non-text portions of this message have been removed]

3 2

uBlas - matrix comparison A == B -> how-to guide ...
by geert_ceuppens 05 May '03

05 May '03

Sharing my effort in working with ublas ... Forgive me if this is described already somewhere - I could not find it immediately in the doc so I looked around in the code ; ) Matrix comparison is implemented using the equals() method defined in matrix_assign.hpp: bool equals (const matrix_expression<E1> &e1, const matrix_expression<E2> &e2) Code snippet: ublas::matrix<double> A(5,5), B(5,5); // Initialise A somehow init_matrix(A); B = A; // This does not compile ... bool equal = ( A == B ); // But this does (and it works ...) bool equal = ublas::equals ( A, B );

1 0

RE: [Boost-Users] How To Build Boost 1.30 With Dev-CPP?
by bjorn.karlsson＠readsoft.com 04 May '03

04 May '03

> From: firingme [mailto:yg-boost-users@m.gmane.org] > E:\LY\boost_1_30_0>bjam sTOOS=mingw This should read sTOOLS, not sTOOS. Bjorn

5 5

RE: [Boost-Users] Code Fail
by Jim.Hyslop 02 May '03

02 May '03

Hossein Haeri [mailto:powerprogman@yahoo.com] wrote: > There is a tiny cpp file attached to this message > that raises the following exception while running: No, there isn't ;-) The email list strips out attachments. If it truly is tiny, then paste it inline. If it's not-so-tiny, then you can upload it to the files area: http://groups.yahoo.com/group/Boost-Users/files/ -- Jim

1 0

RE: [Boost-Users] Re: Is there a "known issues" page anywhere?
by Jim.Hyslop 02 May '03

02 May '03

Alisdair Meredith [mailto:yg-boost-users@gmane.org] wrote: > "Jim.Hyslop" wrote: > > > There is a "Known Issues" section on the wiki (see the link in the > > auto-appended .sig) which is intended to document these > common "gotchas". I > > haven't looked at it for a while, so I don't know how > up-to-date it is. > > I tried looking there first, and after your message went back for a > second look. > Perhaps I am being dense, but I really can't find it. > Closest match is > the 'build issues' page, which I think is for a different thing. > 'Search' the wiki did not help either. That's the page I meant - CommonBuildProblems: http://www.crystalclearsoftware.com/cgi-bin/boost_wiki/wiki.pl?CommonBuildPr oblems Sorry about the confusion. Your problem with the lexical_cast is the precise reason I created the page. > Could you post a link to the precise page you were thinking > about? And > I think a more prominent link on the front page may be useful too. There's one under the "Quick Links" section, but as you say it's not very prominent. Remember, the point of a wiki is that *anyone* can change the information. If you think it deserves a more prominent position, then you are free to give it one :=) -- Jim

1 0

PP library...
by Eric Robert 02 May '03

02 May '03

Hi, I think the preprocessor library could help me here but I'm having problems understanding it. Maybe someone will be so kind to explain it to me a bit. Here is what I want to produce: void foo(char); void foo(int); void foo(unsigned int); .... You got the idea. Something that repeats a list of things i.e. here, types. How can I do that? Thanks! Eric [Non-text portions of this message have been removed]

3 3

Code Fail
by Hossein Haeri 02 May '03

02 May '03

Dear all, There is a tiny cpp file attached to this message that raises the following exception while running: "Project ConstStatProject.exe raised exception class EAccessViolation with message 'Access violation at adress ...'. Process stopped. Use Step or Run to continue." and stops at the following line of the share_ptr library: if(use_count_ == 0 && weak_count_ != 0) boost::throw_exception(boost::bad_weak_ptr()); Can anybody help me? BTW, I'm using BCB6. Thanks, --Hossein __________________________________ Do you Yahoo!? The New Yahoo! Search - Faster. Easier. Bingo. http://search.yahoo.com [Non-text portions of this message have been removed]

1 0

[BGL] Online Graph Algorithms (fwd)
by David Pearce 02 May '03

02 May '03

Hello, I am interested in how boosters think online graph algorithms should be written. I have looked at the connected_components algorithm and feel that it breaks some of the ideas behind generic programming. For example, consider the following: > typedef adjacency_list <vecS, vecS, undirectedS> Graph; > Graph G(6); > add_edge(0, 1, G); > add_edge(1, 4, G); > > // create the disjoint-sets structure, which requires > // rank and parent vertex properties > ... > initialize_incremental_components(G, ds); > incremental_components(G, ds); > > generic_algorithm_which_adds_edges(G); At this point, the connected_components data structures may be out-of-sync. Now, I suppose that we could write a wrapper class which looked like a "Graph" and provides its own add_edge to update "ds" correctly. However, this seems overly cumbersome. It strikes me that a better way of doing this would be to use an abstract base class, like so: > template<class T, ...> > class connected_components : public T { > ... > }; > > template<class T> > pair<T::edge_descriptor, bool> > add_edge(connected_components<T>::vertex_descriptor t, ...) { > // do stuff > return add_edge(static_cast<T::vertex_descriptor>(t), ...); > } Thus, we supply "Graph" as the base class and can then safely pass it to "generic_algorithm_which...". So, does anyone know of any reasons not to do this? Cheers, David J. Pearce

1 0