- Boost-users - lists.boost.org

about shared_ptr
by Daniel Oberhoff 28 Jun '07

28 Jun '07

Hi all, I am using shared_ptr to ease implementation of a matlab file wrapper. Esepcially the fact that it is in tr1 and thus available on many platforms already pushed that decision. Now I was wondering about one thing: Is there a devent way, using shared_ptr, to hand out multiple references? What I mean is something like this: class Variable { shared_ptr<Variable> ref(); }; The problem with shared_ptr is, that, or so it seems, you can only create one directly from the object like this: shared_ptr<Variable> p(new Variable(...)); and all further references have to be created as copies of this (or copies of copies...). So I have to keep one of those around to make new ones. But I cannot keep one inside the class itself or else it will never be deleted (because it would reference itself). The same goes for weak_ptr, since there must be a live shared_ptr around for the object not to be deleted. The only way I can think of is implementing my own refcount and hooking into my own release method. THe problem with this is that I cannot, acoording to the docs, hook up the retain method, so each shared_ptr "cloud" (i.e. set of shared_ptrs that have been created from a common shared_ptr) is hooked up to one ref_count, so they might invalidate when the object stays alive, making any attached weak_ptrs useless. The other is using intrusive_ptr, but that a) is not in tr1 and b) does not have a weak ptr. I was also wondering why intrusive_ptr is not wrapped inside shared_ptr, i.e. why I cannot hook up both retain (increase refcount) and release into shared_ptr, but I suppose that is for the developers list. One other way I could think of would be that I keep a reference to the global ref_count of all the shared_ptrs, and creating new shared_ptrs on top of that, thus binding all shared_ptrs together, but that part does not seem to be exposed in the api. Any Ideas? Daniel Oberhoff ps: should it be possible to post to this list via gmane? I would like that much better, so this won't clog my inbox. But trying to post I keep getting error 441 in gnus

3 2

Re: [Boost-users] [serialization] problem with derived classes
by Johan Franzén 28 Jun '07

28 Jun '07

> Hello, > > I'm having an issue with serializing derived classes from a base class > pointer. I think I've narrowed it down to the simplest possible code. I > get an "unregistered class" exception at runtime even though I've > registered the derived class with BOOST_CLASS_EXPORT. Here's the code: > [ snip ] > ----------------------------------------------- > > X.hpp: > > #pragma once > > #include <boost/serialization/base_object.hpp> > #include "A.hpp" > > class X : public A > { > friend class boost::serialization::access; > template<class Archive> > void serialize(Archive & ar, const unsigned int) > { > ar & boost::serialization::base_object<A>(*this); > } > }; > > ------------------------------------------------ > > X.cpp: > > #include "X.hpp" > > #include <boost/serialization/export.hpp> > > BOOST_CLASS_EXPORT(X) > > ----------------------------------------------- > > As you can see, I don't have BOOST_CLASS_EXPORT in the X.hpp header. > This is the way I originally had it but then I got multiple definition > link errors when more than one class derived from A. The suggestion in > the serialization docs was to move the macro to a .cpp file. [ snip ] Hello Avery, I suspect You need to BOOST_CLASS_EXPORT(A) as well, since You do try to serialize an A in X's serialize() method. This is was I ended up doing when serializing a hierarchy of derived classes and a base class. I'm not sure how to do that in Your case, since I had all classes and all BOOST_CLASS_EXPORT macros in a single header file. HTH, Johan Franzén

4 4

Re: [Boost-users] [multi-index /mpl] fata l compile error when using a tag in VS7.1 with SP1
by hongleij＠126.com 28 Jun '07

28 Jun '07

when i clean the solution and then rebuild it,it is useless; >* Try disabling the option /Gm (Enable Minimal Rebuild.) Yes,i just turn off it,then the error disappear. ps: its really a painful thing waiting for compiling, it about 10s for that litte file in a pentium 2.8G cpu. 一起来，150 万人同时在玩的梦幻西游

2 1

Creating a property map inside Dijkstra Visitor
by Jonas Gabriel 28 Jun '07

28 Jun '07

Good afternoon, Inside a dijkstra visitor I would like to change the edge weights at some point, so in my visitor I'm trying to make a property map called weight : > template<class DynamicMap> > class gds_visitor : public dijkstra_visitor<> > { > public: > > gds_visitor(DynamicMap input):dynamic(input){} > > .... > > template <class Edge, class Graph> > void examine_edge (Edge e, Graph& g) { > typename property_map<Graph, edge_weight_t>::type weight; > weight=get(edge_weight, g); > > std::cout << "An edge is being examined..." << "\n"; > } > > protected: > DynamicMap dynamic; > }; where Graph: > typedef adjacency_list<listS, vecS, directedS, no_property, property<edge_weight_t, int> > Graph; and Edge: > typedef std::pair<int, int> Edge; But it is not compiling, mentioning a problem with weight 'soperator = . Thank you in advance, Jonas Gabriel

2 1

grapg problem: getting all vertices between source and sink
by gast128 28 Jun '07

28 Jun '07

Dear all, we have a simple problem: we want to know all vertices between a source and sink vertex in adirected graph. So my thoughts were to see if a vertex is reachable from both source and sink (by doing a breadth_first_search on the graph from source and a breadth_first_search from the sink on the reverse graph) and if reachable they lay in between. However this counts too much vertices. Consider graph So -> Si -> C. C is reachable from both So as reachable from Si using in edges. This is not good. The breadth_first_search should not perform an out_edge discover from Si. Do I have to implement my own breadth_first_search_with_stop or are there alternatives? Wkr, me

2 3

[multi-index /mpl] fatal co mpile error when using a tag in VS7.1 wit h SP1
by hongleij＠126.com 28 Jun '07

28 Jun '07

//boost #include <boost/multi_index_container.hpp> #include <boost/multi_index/hashed_index.hpp> #include <boost/multi_index/member.hpp> #include "boost/multi_index/key_extractors.hpp" using namespace ::boost; using namespace ::boost::multi_index; #include <string> struct Peer { std::string userid; int age()const {return age_;} private: int age_; }; struct CPeer:public Peer { int index; }; struct Peer_age_extractor { typedef int result_type; const int operator()(const CPeer& peer)const { return peer.age(); } }; struct userid{}; struct age{}; typedef multi_index_container< CPeer, indexed_by< hashed_unique<tag<userid>,member<Peer,std::string,&Peer::userid> >, hashed_non_unique<tag<age>,const_mem_fun<Peer,int,&Peer::age> >// // hashed_non_unique<const_mem_fun<Peer,int,&Peer::age> > //without a tag will be ok. //hashed_non_unique<tag<age>,Peer_age_extractor > //this line with cause the same question. > > CPeersType; I get this in vs7.1 with sp1: ------ Build started: Project: multi_index1, Configuration: Debug Win32 ------ Compiling... hash_test.cpp f:\boost_1_34_0\boost\mpl\aux_\preprocessed\plain\apply_wrap.hpp(48) : fatal error C1001: INTERNAL COMPILER ERROR (compiler file 'f:\vs70builds\6030\vc\Compiler\CxxFE\sl\P1\C\main.c', line 1164) Please choose the Technical Support command on the Visual C++ Help menu, or open the Technical Support help file for more information Build log was saved at "file://g:\cpp\boost\multi_index1\Debug\BuildLog.htm" multi_index1 - 1 error(s), 0 warning(s) yet this can compile well in gcc 3.4. i really don't know why. many thanks for JOAQUIN LOPEZ MU?Z 's past help.;-) 百万玩家同玩的乐园，人气爆发的梦幻西游

2 1

Passing variable no. of arguments in macro
by Prasanta K Chakravarty 28 Jun '07

28 Jun '07

Hi, I have a requirement of passing variable no. of arguments in macro. So, I have build Boost wave library and got .dll and .lib file. But I am not able to use it. Please help me out how to use that one. Thanks & Regards, Prasanta

1 0

Re: [Boost-users] [serialisation]Portable Binary Archives
by Pfligersdorffer, Christian 27 Jun '07

27 Jun '07

boost-users-bounces(a)lists.boost.org wrote: >> If you're talking about portable binary archives for >> Boost.Serialization, there's been talk, but I'm not sure what >> the status is - anyone have an implementation? I'd be willing >> to help / work with someone on it. >> > > That's what I am looking for. We need to get the file sizes down. We > could use text (XML is great for debugging but too big for final > release), but again that has its overheads. We generally don't use > floating point so that may help us. The link I posted does have a > portable implementation which I think uses IEEE754. I just wondered if > there was an official boost:serilisation version; obviously not!! Hello James, there _is_ a portable binary archive provided with the examples but as Robert Ramey points out in the code documentation, it is not complete as it lacks the floating point conversion. But if Johan is willing to fill this gap I am perfectly confident that we could have a comprehensive implementation in a few weeks. That is, if the issue that Dan Leibovich found in the example can be resolved. I follow the portable archive discussion with great interest since I would very much like to use it too. PS @ Cliff: Very interesting post! Do you know about when and if the endian library will enter the official boost release? I would find that useful too. Regards, -- Christian Pfligersdorffer Software Engineering EOS GmbH

6 8

[Serialization] portable binary archive
by Dan Leibovich 27 Jun '07

27 Jun '07

Hi, Few month ago I have started a thread about an issue with the potable binary archive demo: http://lists.boost.org/boost-users/2007/03/26274.php Robert, Were you able to find the exact bug? Thanks in advance Dan

2 2

[serialization] problem with derived classes
by Avery Fay 27 Jun '07

27 Jun '07

Hello, I'm having an issue with serializing derived classes from a base class pointer. I think I've narrowed it down to the simplest possible code. I get an "unregistered class" exception at runtime even though I've registered the derived class with BOOST_CLASS_EXPORT. Here's the code: ---------------------------------------------- main.cpp: #include <boost/archive/text_oarchive.hpp> #include <boost/archive/text_iarchive.hpp> #include <sstream> #include "A.hpp" #include "X.hpp" int main() { std::stringstream ss; const A * const test = new X; { boost::archive::text_oarchive oa(ss); oa << test; } A *test2; { boost::archive::text_iarchive ia(ss); ia >> test2; } return 0; } ----------------------------------------------- A.hpp: #pragma once #include <boost/serialization/access.hpp> class A { private: friend class boost::serialization::access; template<class Archive> void serialize(Archive & ar, const unsigned int) { ar & test; } int test; public: A() : test(0) {} virtual ~A() {} }; ----------------------------------------------- X.hpp: #pragma once #include <boost/serialization/base_object.hpp> #include "A.hpp" class X : public A { friend class boost::serialization::access; template<class Archive> void serialize(Archive & ar, const unsigned int) { ar & boost::serialization::base_object<A>(*this); } }; ------------------------------------------------ X.cpp: #include "X.hpp" #include <boost/serialization/export.hpp> BOOST_CLASS_EXPORT(X) ----------------------------------------------- As you can see, I don't have BOOST_CLASS_EXPORT in the X.hpp header. This is the way I originally had it but then I got multiple definition link errors when more than one class derived from A. The suggestion in the serialization docs was to move the macro to a .cpp file. Full output when I run this: terminate called after throwing an instance of 'boost::archive::archive_exception' what(): unregistered class Aborted (core dumped) Any help would be appreciated greatly, Avery

1 0