September 2005 - Boost-users - lists.preview.boost.org

Memory leak in zlib iostreams unit
by Will Bryant 24 Sep '05

24 Sep '05

Hi all/Jonathan, valgrinding my applications that use zlib reveals that the zlib filter classes leak memory on destruction, as the call to inflateInit2/deflateInit2 is not matched by a corresponding call to inflateEnd/deflateEnd. I notice that boost/iostreams/filter/zlib.hpp has: template<typename Alloc> zlib_compressor_impl<Alloc>::~zlib_compressor_impl() { /*reset(true, false);*/ } and: template<typename Alloc> zlib_decompressor_impl<Alloc>::~zlib_decompressor_impl() { /*reset(false, false);*/ } - which I imagine were added in response George M. Garner's note about what I assume was the same problem (http://lists.boost.org/Archives/boost/2005/03/83379.php) Uncommenting those calls to reset fixes the memory leaks. Are they meant to be commented out in the release? Cheers, Will -- Will Bryant http://carcino.gen.nz/ will(a)core-dev.co.nz +64 21 655 443

2 1

mpl question
by Karl Meerbergen 24 Sep '05

24 Sep '05

Dear boost users, I have a question regarding mpl. Consider the following piece of code: typedef boost::mpl::vector<int,double> v_i_d_t ; typedef boost::mpl::pop_front<v_i_d_t>::type v_d_t ; Is v_d_t equal to boost::mpl::vector<double> ? I do not have the impression. Thanks, Karl

3 6

Error populating pointer container with stack based objects
by Allen Cronce 23 Sep '05

23 Sep '05

1 0

Re: [Boost-users] Python 'in' keyword-like in C++
by Foster Gareth 23 Sep '05

23 Sep '05

What do people think of actually including things like this in code? In terms of maintenance and readability, is it a good idea to create so many different ways of doing things? Purely thinking from the POV that, I find a library I want to modify to support some new thing, and I find it is full of unusual styles and obfuscated custom techniques. I'm not knocking boost or the stl, just wondering if things like trying to recreate python styles in C++ are a step too far, a sort of "what is wrong with the C++ style?" question. No flames, I'm _not_ telling anybody they are 'wrong' :) Gaz

3 2

[Serialization] Serializing classes defined in DLLs
by Topi Mäenpää 23 Sep '05

23 Sep '05

Hi, How do I register classes that are defined in a dynamically loaded so/dll? Using BOOST_CLASS_EXPORT just causes "terminate called after throwing an instance of 'boost::archive::archive_exception' what(): unregistered void cast". My set-up is as follows: class A { public: A(); virtual ~A() {} private: // polymorphic archive stuff here }; class B : public A {} BOOST_CLASS_EXPORT(B); Serializing instances of B through a A* works fine if B is defined in the main executable. Now, I have tried the following (approximately): // An exported function that returns a void pointer typedef void* (*func)(); int main() { // Load a dynamic library dll lib = load_dll("mylib.dll"); // Get a pointer to the instance creator function func instance_creator = (func)lib.resolve("instance_creator"); // An instance of B is created in the dll and returned as a void*. A* a = (A*)instance_creator(); ... archive & a; //throws unregistered void cast } B is both defined and exported in mylib.dll. So, what is the deal? I don't understand the soul of BOOST_CLASS_EXPORT well enough to solve the problem myself. But since I'm using the dll version of the serialization library, I'd assume that the dynamically loaded dll (mylib.dll) and the main executable use the same address space for the archive classes. I would understand this if I encountered this problem with static linking because I'd need to inlude two copies of the serialization library... Any advice? Regards, Topi

3 11

Python list comprehension-like syntax in C++
by Tzu-Chien Chiu 23 Sep '05

23 Sep '05

In Pyhon: y = [foo(_1) for _1 in x] In C++: list<int> x; x.push_back(...); .. list<int> y; for (list<int>::iterator it = x.begin(); x.end() != it; ++ it) y.push_back( foo(y) ); With boost, can we write the Python-like statement: copy( x.begin(), x.end(), magic_back_inserter(foo(_1), y) ); // pseudo-code -- Tzu-Chien Chiu, 3D Graphics Hardware Architect <URL:http://www.csie.nctu.edu.tw/~jwchiu>

5 6

example with Matrix iterator1, iterator2
by DrJekyll 23 Sep '05

23 Sep '05

dear group members I read through the Ublas document but could not see any example with the iterator1 and iterator2 of matrix. Could someone kindly send me some example and explanation please? thanks XS ________________________________________________________________ Sent via the WebMail system at empirics.net

1 0

[BBv2] getting started problem with headers
by Tino S. 23 Sep '05

23 Sep '05

Hi boosters, i can't build my first stand alone project with boost.build. I did'nt found any solutions in the doc. :( Here the facts: envirement: - winxp with vc7.1 and cygwin with gcc3.4 - bjam version 03.01.11 (which comes with boost 1.33) - set the SystemVar BOOST_ROOT=D:\Temp\boost_1_33_0 (under win) In my home directory are these files: // FILE: site-config.jam # Get Boost location from build system path-constant BOOST_ROOT : [ modules.peek : BOOST_ROOT ] ; // FILE: user-config.jam import toolset : using ; using msvc ; and here is my simple project dir + my_prj // FILE: hello.cpp #include <boost/array.hpp> #include <iostream> int main() { boost::array<int,4> a = { { 1, 2, 3 } }; std::cout << "Hello!\n"; std::cout << "a[1]:" << a[1] << '\n'; std::cin.get(); return 1; } // FILE: Jamfile # nothing works, :( .. could not find the boost header # nothing works.. exe hello : hello.cpp ; #exe hello1 : hello.cpp : <include>boost ; #exe hello2 : hello.cpp : <include>BOOST_ROOT ; #exe hello3 : hello.cpp : <include>$(BOOST_ROOT) ; // FILE: project-root.jam path-constant MY_DIR : . ; using lex ; using bison ; using testing ; What's wrong? Why could'nt found the boost header files? For beginners it would wounderfull to add samples to the doc which shows the integration of Boost (include and libs). Thanks in advance Tino Schumacher ___________________________________________________________ Gesendet von Yahoo! Mail - Jetzt mit 1GB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de

2 1

Re: [Boost-users] [BGL] detect cycles in a directed graph
by Andrea Olgiati 23 Sep '05

23 Sep '05

Giulio, Here's an idea. You say you want to list all the vertices that make up any one of the (potentially very numerours) cycles that originate on a vertex X. If _any_ cycle will do, how about the shortest? You can use Dijkstra. From http://www.boost.org/libs/graph/doc/dijkstra_shortest_paths.html: "Also you can record the shortest paths tree in a predecessor map: for each vertex u in V, p[u] will be the predecessor of u in the shortest paths tree (unless p[u] = u, in which case u is either the source or a vertex unreachable from the source)." I don't think Dijkstra will work on a path that looks like V->X->Y->U->V (ie, a cycle). But, to get around this, you do know that U->V is an edge, so you can compute the shortest path between V and U, and know that there's an edge between U and V. HTH, Andrea -- Andrea Olgiati - Elixent - Castlemead, Lwr Castle St., Bristol BS1 3AG, UK andrea.olgiati(a)elixent.com ++44 (0)117 9175612 > -----Original Message----- > From: boost-users-bounces(a)lists.boost.org > [mailto:boost-users-bounces@lists.boost.org] On Behalf Of > Giulio Veronesi > Sent: 22 September 2005 11:05 > Cc: boost-users(a)lists.boost.org > Subject: Re: [Boost-users] [BGL] detect cycles in a directed graph > > Thanks Vladimir. > > I'm not sure that I've followed well your description. Here > is my test implementation... but it doesn't work. > > Please, could you help me to discover the problem? > > Thanks in advance, > Giulio > > > typedef adjacency_list<vecS, vecS, directedS, > no_property, no_property> MyGraph; > > typedef graph_traits<MyGraph>::vertex_descriptor Vertex; > > > struct CycleDetector : public dfs_visitor<> { > CycleDetector(std::vector<Vertex> p, > std::vector<Vertex>& _cycle) : > m_predecessor(p), cycle(_cycle) { } > > template <class Edge, class Graph> > void back_edge(Edge e, Graph& g) > { > Vertex t = target(e, g); > Vertex c = source(e, g); > > cycle.push_back(c); > do { > c = m_predecessor[c]; > cycle.push_back(c); > } while(c != t); > } > > protected: > std::vector<Vertex>& cycle; > std::vector<Vertex> m_predecessor; > }; > > int main(int,char*[]) > { > > typedef std::pair<int,int> Edge; > std::vector<Edge> edges; > edges.push_back(Edge(0,1)); > edges.push_back(Edge(1,2)); > edges.push_back(Edge(2,3)); > edges.push_back(Edge(3,1)); > edges.push_back(Edge(3,4)); > edges.push_back(Edge(4,0)); > > MyGraph g(edges.begin(), edges.end(), 5); > > std::vector<Vertex> cycle; > std::vector<Vertex> p(num_vertices(g)); > CycleDetector vis(p, cycle); > depth_first_search(g, visitor(vis)); > > for (int i=0; i<cycle.size(); i++) > std::cout << cycle[i] << ", "; > std::cout << std::endl; > > return 0; > } > > > Vladimir Prus ha scritto: > >>something like this? > >> > >>template <class Edge, class Graph> > >>void back_edge(Edge e, Graph& g) { > >> clinks.push_back(TLink(source(e, g), target(e, g))); } > >> > >>where: > >> > >>typedef std::pair<int,int> TLink; > >>vector<TLink> clinks; > > > > > > Something more like: > > > > struct vis > > { > > vis(vector_property_map<int> predecessor) > > : m_predecessor(predecessor) {} > > > > template<......> > > void back_edge(Edge e, Graph& g) > > { > > vertex_descriptor t = target(e, g); > > vertex_description c = source(e, g); > > vector<vertex_descriptor> cycle; > > cycle.push_back(c); > > do { > > c = m_predecessor[c]; > > cycle.push_back(c); > > } while(c != t); > > } > > }; > > > > On construction, you need to pass the same > vector_property_map both to > > the visitor and the depth_first_search algorithm. > > > > Yes, this is untested code, sorry :-( > > > > - Volodya > > > > > > _______________________________________________ > Boost-users mailing list > Boost-users(a)lists.boost.org > http://lists.boost.org/mailman/listinfo.cgi/boost-users > >

2 1

Re: [Boost-users] A question about boost::lambda::if_ behavior
by tom gee 23 Sep '05

23 Sep '05

Thanks to Simon for the elaborate explanation. Thanks to Bjorn again for the url, getting me to the right place. Yes, I should have gone through the docs. (*^�n^*) > Message: 1 > Date: Thu, 22 Sep 2005 20:44:08 +1200 > From: Simon Buchan <simon(a)hand-multimedia.co.nz> > Subject: Re: [Boost-users] A question about boost::lambda::if_ > behavior > To: boost-users(a)lists.boost.org > Message-ID: <dgtqts$e0j$1(a)sea.gmane.org> > Content-Type: text/plain; charset=ISO-8859-1; format=flowed > > Bjorn.Karlsson(a)readsoft.com wrote: > > Hello Tom, > > > > -----Original Message----- > > From: tom gee [mailto:rockonedge@gmail.com] > > Subject: [Boost-users] A question about boost::lambda::if_ behavior > > > > > >>Can anyone please explain this code: > > > > [snip] > > > > The problem is that the expression std::cout << constant("Bingo:<") is > > evaluated immediately. Why? Because it's not a lambda expression. You have > > two options that will help you turn your if-branch into a lambda expression; > > either use boost::lambda::var or boost::lambda::constant. > > > > Using boost::lambda::var: > > > > std::for_each(setFinalVal.begin(),setFinalVal.end(), > > (std::cout << _1 << " ", if_(_1 < 24.000001 && _1 > 23.999999) > > [var(std::cout) << "Bingo:<" << _1 << ">"])); > > > > Using boost::lambda::constant: > > > > std::for_each(setFinalVal.begin(),setFinalVal.end(), > > (std::cout << _1 << " ", if_(_1 < 24.000001 && _1 > 23.999999) > > [std::cout << constant("Bingo:<") << _1 << ">"])); > > > > > > Cheers, > > Bjorn Karlsson > constant should be better (lower overhead, conceptually closer), > but no biggy. I beleive this is the No.1 FAQ, is it not? > (PS: why does not the lambda docs say in the function call overload > section to use BLL's bind? It took me AGES to figure out how to do > that!) > > Message: 7 > Date: Thu, 22 Sep 2005 22:37:28 +1200 > From: Simon Buchan <simon(a)hand-multimedia.co.nz> > Subject: Re: [Boost-users] A question about boost::lambda::if_ > behavior > To: boost-users(a)lists.boost.org > Message-ID: <dgu1ic$2g2$1(a)sea.gmane.org> > Content-Type: text/plain; charset=ISO-8859-1; format=flowed > > tom gee wrote: > > Hello Bjorn, > > > > Thanks for such a quick help. I works now. > > > > Can you please further explain the usage of var and constant . > > in the expression > > if_(_1 < 24.000001 && _1 > 23.999999)[ cout <<"Bingo:<"<<_1<<">"]) > > I understand "Bingo:<" is evaluated immediately, why not ">", which is not a > > lambda expression either, is not > > evaluated along. > > > Refresher course: an expression like: > cout << foo << bar << endl; > is read (aka parsed) by the compiler to be shorthand for the following: > (((cout << foo) << bar) << endl); > which, incidently is just shorthand for: > operator<<(operator<<(operator<<(cout, foo), bar), endl); > Thus, '<<' is known as a 'left assosiative' (I think :D) operator. > Refresher over. > Basicly, the 'cout << "Bingo:<"' sub-expression has no 'lambda stuff' in > it, and therefore calls the _normal_ operator<<, which immediately > prints "Bingo:<". This operator<< returns a reference to cout, so the > next '<<' sees: > cout << _1 > Which (obviously) is an expression containing a lambda (sub-)expression. > The guys who made this library were nice enough to provide overloads of > '<<' (along with every other possible operator) that return a 'lambda > function object', you just have to know that it is a valid lambda > expression that _will_ evalute as "cout << _the_first_argument_" when > called. The same happens with the next <<, it sees: > _lambda_operator_left_shift_ << ">" > , where the left argument is the previous lambda function object made > from the "cout << _1" sub-expression. Now there is _another_ overload of > << that creates _another_ lambda function object, that _will_ evalute as > if it was: > cout << _the_first_argument_ << ">" > So, basicly, anytime there is a sub-expression that doesn't have a > lambda object in it, it is treated as a normal expression, but if it > does, it creates a function object that will do what it looks like it > would do. What 'constant' and 'var' do is make a (lambda?) function > object that just returns it's argument, as, respectively, a constant > copy, or a reference to the original (I think, again, use boost::ref to > be sure) > > Forgive me if this question seems stupid, but I am new to Lambda, fascinated > > by its power, but convoluted by its mechanism yet. > > > > Thanks. > > > [snip] > > Don't worry about it, but don't forget to read the documentation! > One more thing to keep in mind, the overload for '()', the call > operator, for a lambda function object immeadiatly evaluates it's > contained expression, ie: > (_1 + 3)(4) == (4 + 3); > to get around this if you want to delay a call to that object, use: > boost::lambda::bind(_1 + 3, 4); > which returns a lambda function object that when called, calls the first > argument with the rest of bind's arguments as it's arguments, so: > bind(_1 + 3, 4)() == (_1 + 3)(4) == (4 + 3) > Which is VERY usefull if you want to, say call functions in a STL container: > > int i, j; > std::vector<void(*)(int, int)> some_funcs; > ... > std::for_each(some_funcs.begin(), some_funcs.end(), bind(*_1, i, j)); > > Although, Boost.Signals is probably a better bet if you do stuff like > that a bit. >

1 0