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

Re: [Boost-users] boost::lambda::bind question
by Vishnevetsky, Eugene (IT) 07 Jul '05

07 Jul '05

I came across this problem in the real project where I don't control the class that corresponds to class Y of the example, so I can't change the signature of the member function. I could write a wrapper function that takes Y*, but after some point the code with a simple 'for' loop becomes cleaner and more readable than the one using lambda. If this problem is not a bug, but an inherent limitation of the lambda library or its current implementation, it would be nice if it is mentioned in the documentation. Thanks, Eugene Vishnevetsky Roman Krylov wrote: Perhaps the only solution here is to use Y* instead of Y& with corresponding modifications; this way there is no need for var() or smth similar. Why not? Vishnevetsky, Eugene (IT) wrote: >Is it possible to use boost::ref() or boost::lambda::var() of an >abstract class as a parameter to boost::lambda::bind()? >It seems that compiler attempts to instantiate an object of that >abstract class. >Example: > >----------------- >using namespace boost; >using namespace boost::lambda; >class Y >{ >public: > virtual void foo() = 0; > void foo2 (){ } >}; >class Z : public Y >{ >public: > void foo(){} >}; > >class TT >{ >public: > void test( Y& y ) { y.foo2(); } >}; > >int main() >{ > std::vector<TT> xt; > Z z; > Y &y = z; > std::for_each (xt.begin(), xt.end(), bind(&TT::test,_1, >ref(y))); > return 0; >} >---------------------- > >And Error message from g++ (GCC) 3.2.3: > >--------------------------------- >/boost/tuple/detail/tuple_basic.hpp: In instantiation of >`boost::tuples::cons<Y, boost::tuples::null_type>': >/boost/tuple/detail/tuple_basic.hpp:326: instantiated from >`boost::tuples::cons<boost::tuples::null_type, boost::tuples::cons<Y, >boost::tuples::null_type> >' >/boost/tuple/detail/tuple_basic.hpp:326: instantiated from >`boost::tuples::cons<void (TT::*const)(Y&), >boost::tuples::cons<boost::tuples::null_type, boost::tuples::cons<Y, >boost::tuples::null_type> > >' >/boost/lambda/detail/lambda_functor_base.hpp:151: instantiated from >`boost::lambda::detail::has_null_type<boost::tuples::cons<void >(TT::*const)(Y&), boost::tuples::cons<boost::tuples::null_type, >boost::tuples::cons<Y, boost::tuples::null_type> > > >' >/boost/lambda/detail/lambda_functor_base.hpp:229: instantiated from >`boost::lambda::detail::deduce_non_ref_argument_types<boost::tuples::tu >p >le<void (TT::*const)(Y&), const >boost::lambda::lambda_functor<boost::lambda::placeholder<1> >, Y&, >boost::tuples::null_type, boost::tuples::null_type, >boost::tuples::null_type, boost::tuples::null_type, >boost::tuples::null_type, boost::tuples::null_type, >boost::tuples::null_type>, boost::tuples::null_type>' >/boost/lambda/detail/lambda_functor_base.hpp:412: instantiated from >`boost::lambda::lambda_functor_base<boost::lambda::action<3, >boost::lambda::function_action<3, boost::lambda::detail::unspecified> >>, boost::tuples::tuple<void (TT::*const)(Y&), const >boost::lambda::lambda_functor<boost::lambda::placeholder<1> >, Y&, >boost::tuples::null_type, boost::tuples::null_type, >boost::tuples::null_type, boost::tuples::null_type, >boost::tuples::null_type, boost::tuples::null_type, >boost::tuples::null_type> >::sig<boost::tuples::null_type>' >/boost/lambda/detail/lambda_functors.hpp:135: instantiated from >`boost::lambda::lambda_functor<boost::lambda::lambda_functor_base<boost : >:lambda::action<3, boost::lambda::function_action<3, >boost::lambda::detail::unspecified> >, boost::tuples::tuple<void >(TT::*const)(Y&), const >boost::lambda::lambda_functor<boost::lambda::placeholder<1> >, Y&, >boost::tuples::null_type, boost::tuples::null_type, >boost::tuples::null_type, boost::tuples::null_type, >boost::tuples::null_type, boost::tuples::null_type, >boost::tuples::null_type> > >' >lamtest.cpp:35: instantiated from here >/boost/tuple/detail/tuple_basic.hpp:320: cannot declare field >`boost::tuples::cons<Y, boost::tuples::null_type>::head' >to be >of type `Y' >/boost/tuple/detail/tuple_basic.hpp:320: >because the following virtual functions are abstract: >lamtest.cpp:12: virtual void Y::foo() >---------------------------------------- > >Thanks, > >Eugene Vishnevetsky >-------------------------------------------------------- > >NOTICE: If received in error, please destroy and notify sender. Sender does not waive confidentiality or privilege, and use is prohibited. > >_______________________________________________ >Boost-users mailing list >Boost-users(a)lists.boost.org >http://lists.boost.org/mailman/listinfo.cgi/boost-users <http://lists.boost.org/mailman/listinfo.cgi/boost-users> > > > -------------------------------------------------------- NOTICE: If received in error, please destroy and notify sender. Sender does not waive confidentiality or privilege, and use is prohibited.

4 5

johnson_all_pairs_shortest_paths under MSVC 6.0?
by Vivid Yang 07 Jul '05

07 Jul '05

Hi, Does anybody has experience in applying johnson_all_pairs_shortest_paths under WINDOWS? Examples would help me a lot. I have difficulties in compiling the johnson_all_pairs_shortest_paths example in the library: example/johnson-eg.cpp The code D[id2[*u]-1][id2[*v]-1] = get(d, *v) + get(h, *v) - get(h, *u); has error: "subscript requires array or pointer type" Thanks a lot. Vivid

2 1

[signals] Disconnecting slots
by Staal, J.J. (Joes) 07 Jul '05

07 Jul '05

In the docs for signals, an example is given for disconnecting slots: void foo(); void bar(); signal<void()> sig; sig.connect(&foo); sig.connect(&bar); // disconnects foo, but not bar sig.disconnect(&foo); When I try this with MSV 7.1, I get an compiler error: error C2664: 'boost::signal0<R,Combiner,Group,GroupCompare,SlotFunction>::disconnect' : cannot convert parameter 1 from 'void (__cdecl *)(void)' to 'const boost::signal0<R,Combiner,Group,GroupCompare,SlotFunction>::group_type &' with [ R=boost::detail::function_traits_helper<boost::add_pointer<void (void)>::type>::result_type, Combiner=boost::last_value<void>, Group=int, GroupCompare=std::less<int>, SlotFunction=boost::function<void (void)> ] and [ R=boost::detail::function_traits_helper<boost::add_pointer<void (void)>::type>::result_type, Combiner=boost::last_value<void>, Group=int, GroupCompare=std::less<int>, SlotFunction=boost::function<void (void)> ] My question is, are the docs incorrect, or am I doing something wrong? Regards, Joes. This e-mail and its contents are subject to the DISCLAIMER at http://www.tno.nl/disclaimer/email.html

1 0

boost::variant and SGI's compose1 cause constness error
by junkmail.hoefling＠arcor.de 07 Jul '05

07 Jul '05

Hi there, I have a problem using boost::variant. I want apply_visitor to go through a list of pairs whose first type is a variant. Using compose1 and select1st from the SGI extensions to the STL yields a constness error. The problem is illustrated by the attached code which shoud be compiled with the g++ (the SGI extensions are here available via __gnu_cxx). What's wrong? Is it my fault using the library or is it a missing const version in apply_visitor_delayed_t()? Thanks a lot for any hints! Best, Erwin

3 2

circular buffer and gcc 4.0.0
by Vincent Torri 07 Jul '05

07 Jul '05

Hello, I need to use gcc 4.0 and circular buffer. But this latter seems unusuable with gcc 4.0. It gives a lots of error messages. Here is the 2 irst error messages: in debug.hpp: /home/torri/local/include/boost-1_33/boost/circular_buffer/debug.hpp:60: error: invalid use of undefined type 'const struct boost::cb_details::cb_iterator_base' /home/torri/local/include/boost-1_33/boost/circular_buffer/debug.hpp:22: error: forward declaration of 'const struct boost::cb_details::cb_iterator_base' the 2 classes in debug.hpp are cross-referenced. It seems that the forward declaration is not sufficient for gcc 4.0. in base.hpp: /home/torri/local/include/boost-1_33/boost/circular_buffer/base.hpp:139: error: using typedef-name 'boost::circular_buffer<T, Alloc>::iterator' after 'struct' /home/torri/local/include/boost-1_33/boost/circular_buffer/base.hpp:105: error: 'boost::circular_buffer<T, Alloc>::iterator' has a previous declaration here /home/torri/local/include/boost-1_33/boost/circular_buffer/base.hpp:139: error: friend declaration does not name a class or function gcc 4.0 doesn't like the 'struct' word. Putting the 'class' word instead of struct gives error too: /home/torri/local/include/boost-1_33/boost/circular_buffer/base.hpp:139: error: a class-key must be used when declaring a friend /home/torri/local/include/boost-1_33/boost/circular_buffer/base.hpp:139: error: friend declaration does not name a class or function Does someone see solutions for these problems ? Jan: it would be great if you try to finish circular buffer :) thank you Vincent Torri

1 0

BGL - Multiple graphs on the same data-structure.
by sr kumar 07 Jul '05

07 Jul '05

Hi, My application needs to have multiple graphs overlayed on the same set of data. Looks like used bundled properties would be very attractive. Since the data-set is likely to be large and the graphs will have to coexist, I'd like to build graphs of pointers to the base data. However, I am having trouble figuring out how to store pointers to the data structures within an adjacency_list. Any suggestions/pointers to examples would be much appreciated. -sr

2 1

BGL: Brandes betweenness and exterior properties
by Todd A. Gibson 07 Jul '05

07 Jul '05

Hello, I am working with brandes_betweenness_centrality in the BGL and am having difficulty correctly constructing the syntax for retrieving an EdgeCentralityMap using exterior properties. I can successfully invoke the function using bundled properties: struct EdgeAttrs { unsigned int index; double weight; }; template<class GraphType> boost::python::list betweenness_centrality(GraphType& g) { typedef typename GraphType::edge_descriptor Edge; std::vector<float> centmap(num_vertices(g)); std::vector<float> edgecent(num_edges(g)); brandes_betweenness_centrality (g, centrality_map(make_iterator_property_map(centmap.begin(), boost::get(vertex_index, g), float())) .edge_centrality_map(make_iterator_property_map(edgecent.begin(), boost::get(&EdgeAttrs::index, g),float())) .weight_map(boost::get(&EdgeAttrs::weight, g))); ... Among my attempts, here's a typical one that won't compile: typedef typename std::map<Edge, double> EdgeMap; EdgeMap myedges; //... boost::associative_property_map< EdgeMap > edge_assoc_map(myedges); brandes_betweenness_centrality(g, ... .edge_centrality_map(edge_assoc_map) ...) Can someone enlighten me? Thanks, -TAG

2 1

in_edges for a directedS adjacency_list
by sr kumar 07 Jul '05

07 Jul '05

Hi, Is there a reason that the in_edges are not being maintained for a directedS adjacency_list? A quick look through implementations of boost::detail::add_edge(*) indicates correct maintainance for a bidirectional case, and reuse of the out_edges for the undirected case. I'm hopefully not missing something obvious to more enlightened souls and would appreciate any ideas to get the in_edges correctly for a directedS adjacency_list. thanks, -sr

2 1

Re: [Boost-users] Serialization and null pointers
by boost＠iyates.com 07 Jul '05

07 Jul '05

Robert, Thanks for all your help, I really do appreciate it. I was already serializing an stl vector, so you would think I would have thought of that! heh. Anyhow, I read through serialization/shared_ptr.hpp and sort of did the same thing on my custom shared pointer, and it is working great. I especially liked this: #define private public #include <boost/shared_ptr.hpp> #undef private hehe Thanks again! - Chris ----- Original Message ----- From: "Robert Ramey" <ramey(a)rrsd.com> To: boost-users(a)lists.boost.org Date: Wed, 6 Jul 2005 14:48:51 -0700 Subject: Re: [Boost-users] Serialization and null pointers > I suspect you didn't include <boost/serialization/shared_ptr.hpp> which > includes serialization code for shared_ptr. It also includes shared_ptr.hpp > itself. This is similar to the way the headers are organized for all the > stl collections. If you want to serialize the boost share_ptr be sure to > read about in the docs as there are a few gotchas. Also note that there the > "real soon to be released" boost 1.33 has a different shared_ptr and > different serialization of the same. > > Good Luck. > > Robert Ramey > > boost(a)iyates.com wrote: > > I was using my own home-grown one. I wrote it before I started using > > boost though. I decided to try it out with boost::shared_ptr and > > here is what I got: > > > > c:\Boost\boost_1_32_0\boost\serialization\access.hpp(106): error > > C2039: 'serialize' : is not a member of 'boost::shared_ptr<T>' > > > > > > I tried searching on boost shared_ptr serialize but came up empty. > > Any suggestions? Thanks. > > > > ----- Original Message ----- > > From: "Robert Ramey" <ramey(a)rrsd.com> > > To: boost-users(a)lists.boost.org > > Date: Tue, 5 Jul 2005 20:18:00 -0700 > > Subject: Re: [Boost-users] Serialization and null pointers > > > >> boost(a)iyates.com wrote: > >>> Thanks. If I changed the order of the members being serialized it > >>> worked fine. Weird. > >>> > >>> However, I am running into a problem. I am using reference counted > >>> Smart Pointers. If I serialize the reference counter, then when it > >>> is loaded it will likely contain too large of a reference count > >>> because there were more objects referencing the pointer when it was > >>> saved than when it was loaded. However, it works fine it just > >>> doesn't properly delete the objects because the reference count > >>> never reaches 0. > >>> > >>> If I don't serialize the reference count then the objects try to > >>> delete themselves while boost is loading them from the file. I'm > >>> guessing because of the order everything happens, the final > >>> destination isn't increasing the reference count until after boost > >>> already dereferences it. > >>> > >>> Any suggestions? > >> > >> I'm not sure which reference counted Smart Pointer you're using. If > >> you use the boost smart_ptr, its already has serializaiton done. If > >> you want to add serialization to your own Smart Pointer, you can > >> look at how it was done for the boost smart pointer - now there are > >> two versions so you have lots of ideas available. > >> > >> Robert Ramey > >> > >> > >> > >> _______________________________________________ > >> Boost-users mailing list > >> Boost-users(a)lists.boost.org > >> http://lists.boost.org/mailman/listinfo.cgi/boost-users > > > > _______________________________________________ > Boost-users mailing list > Boost-users(a)lists.boost.org > http://lists.boost.org/mailman/listinfo.cgi/boost-users > > >

1 0

Re: [Boost-users] Serialization and null pointers
by boost＠iyates.com 06 Jul '05

06 Jul '05

I was using my own home-grown one. I wrote it before I started using boost though. I decided to try it out with boost::shared_ptr and here is what I got: c:\Boost\boost_1_32_0\boost\serialization\access.hpp(106): error C2039: 'serialize' : is not a member of 'boost::shared_ptr<T>' I tried searching on boost shared_ptr serialize but came up empty. Any suggestions? Thanks. ----- Original Message ----- From: "Robert Ramey" <ramey(a)rrsd.com> To: boost-users(a)lists.boost.org Date: Tue, 5 Jul 2005 20:18:00 -0700 Subject: Re: [Boost-users] Serialization and null pointers > boost(a)iyates.com wrote: > > Thanks. If I changed the order of the members being serialized it > > worked fine. Weird. > > > > However, I am running into a problem. I am using reference counted > > Smart Pointers. If I serialize the reference counter, then when it > > is loaded it will likely contain too large of a reference count > > because there were more objects referencing the pointer when it was > > saved than when it was loaded. However, it works fine it just doesn't > > properly delete the objects because the reference count never reaches > > 0. > > > > If I don't serialize the reference count then the objects try to > > delete themselves while boost is loading them from the file. I'm > > guessing because of the order everything happens, the final > > destination isn't increasing the reference count until after boost > > already dereferences it. > > > > Any suggestions? > > I'm not sure which reference counted Smart Pointer you're using. If you use > the boost smart_ptr, its already has serializaiton done. If you want to add > serialization to your own Smart Pointer, you can look at how it was done for > the boost smart pointer - now there are two versions so you have lots of > ideas available. > > Robert Ramey > > > > _______________________________________________ > Boost-users mailing list > Boost-users(a)lists.boost.org > http://lists.boost.org/mailman/listinfo.cgi/boost-users > > >

2 1