- Boost-users - lists.boost.org

Re: [Boost-users] [Multi-Index] crash on invariant_() violation
by Neil Hunt 03 Apr '09

03 Apr '09

> > ... it doesn't matter if the change is made inside the modify function > as long as modify is called. > > (So, this statement from above does not appear strictly true: > "Key modification has to be done via modify() so as to give the container a > chance to reindex elements.") > > Because (in my case) the change is made elsewhere, I am calling modify with > > a functor that does nothing. > I guess more generally, why is this paradigm implictly disallowed? index_itor->value++; index.update_indices( index_itor ); While I can see the benefit of tying the modification to the multi-index object via modify, there is no explicit reason that I can see why the above is invalid (as per my perfectly working code that uses this paradigm). And "forcing" the user to use a functor to modify can introduce annoying code overhead. Don't get me wrong, I think that using modify with a functor is very clear and helps prevent invalid indices. But, at the same time, it can be a pain to write a functor. (Here's to C++0x lambda's). But in my case, where the code that modifies the object is unaware of multi-indices on that object, it is not feasible to use modify() to make the change. Neil

1 0

[serialization] Private default constructors
by Robert Dailey 03 Apr '09

03 Apr '09

I have a class that can be constructed statically in code, or constructed by data (this is where boost::serialization comes in). For example: class foo { public: foo() {} foo( std::string text ) : m_text( text ) {} private: template< class Archive > void serialize( Archive& archive, unsigned int version ) { archive & m_text; } std::string m_text; friend class boost::serialization::access; }; There is one problem with this, though. It is possible for a user of this class to create a "foo" with the default constructor, thus enabling them to have "zombie" objects. The only thing I want using the default constructor is boost::serialization, not the user. Is there a way I can make the default constructor private and still have boost::serialization access it? Keep in mind that this must work polymorphically too.

1 1

[repost][filesystem][system] Linking errors with C++0X options on intel C++ 11.0
by jesseperla＠gmail.com 02 Apr '09

02 Apr '09

Sorry to repost, but I am at a loss and was hoping I have missed something. I am on Intel C++ 11.0 on windows. I built boost::filesystem and boost::system as static libraries with the following line in my userconfig.jam using intel-win : 11.0 : : <cxxflags>/Qstd:c++0x ; I am then compiling a project (that works with non-C++0x just fine) with /Qstd:c++0x in the command line. But when I try to link it with my calls to boost::filesystem header functions, I get the following link errors (it must be using auto linking) 1>etk_template.obj : error LNK2019: unresolved external symbol "class boost::system::error_category const & __cdecl boost::system::get_system_category(void)" (?get_system_category@system@boost@@YAABVerror_category@12@XZ) referenced in function _main 1>etk_template2.obj : error LNK2001: unresolved external symbol "class boost::system::error_category const & __cdecl boost::system::get_system_category(void)" (?get_system_category@system@boost@@YAABVerror_category@12@XZ) 1>etk_template.obj : error LNK2019: unresolved external symbol "class boost::system::error_category const & __cdecl boost::system::get_generic_category(void)" (?get_generic_category@system@boost@@YAABVerror_category@12@XZ) referenced in function _main 1>etk_template2.obj : error LNK2001: unresolved external symbol "class boost::system::error_category const & __cdecl boost::system::get_generic_category(void)" (?get_generic_category@system@boost@@YAABVerror_category@12@XZ) I am not changing any pre-compiler definitions for boost when I use /Qstd:c++0x. Should I be changing anything there? Am I out of luck? -Jesse

4 3

[BGL] A bug in the edge_range() function.
by Michael Olea 02 Apr '09

02 Apr '09

In boost 1.37.0 there is a "metaprogramming bug" in the edge_range() function: it fails to compile when the Directed Selector template argument to adjacency_list is set to directedS, but does work if the argument is either undirectedS, or bidirectionalS. Attached below is a short program illustrating the problem. The problem is that edge_range() instantiates a StoredEdge via a constructor taking 3 arguments, but when the Directed Selector is directedS StoredEdge is typedef'ed to stored_edge_property<vertex_descriptor, EdgeProperty>, which has no such constructor. One solution might be to create overloaded edge_range_dispatch() functions, and dispatch on Config::on_edge_storage. -- Michael

1 0

[serialization] Serializing "int" types...
by Robert Dailey 02 Apr '09

02 Apr '09

If I choose to serialize a variable with type "int", is this portable? For example, if I stream out this value on a 64-bit system, and then load it in on a 32-bit system, will this result in a buffer-overrun or anything else weird? I read the serialization docs but didn't see this topic in there. Sorry if I missed it. Help is appreciated. Thank you.

2 2

[asio] async_accept shares port?
by Roman Shmelev 02 Apr '09

02 Apr '09

HI! I've got another strange question, hope it is not very dumb. I'm launching two programs on my computer. Each tcp_acceptor in each program calls async_accept on one port and so is ready to accept connections. When I try to connect to the predefined port, I am connecting to the app that called async_accept later. As I thought, when trying to listen to port that is already accepting connections - another program will receive some error_code, but I do not see this in my case. Where is the mistake? Thanks!

1 0

[boost-users] lexical-cast throws exceptions
by Roman Shmelev 02 Apr '09

02 Apr '09

Please take a look at the simple piece of code. When it is compiled on MSVC 2008 (no sp) , Win Sp2 then we receive access violation on program exit Can you tell me the reason of such strange behaviour? (boost 1.38.0) #include "stdafx.h" #include <iostream> #include <boost/lexical_cast.hpp> //#include <boost/thread.hpp> using namespace std; class ctest { public: ~ctest () { boost::lexical_cast<string>(1); } }; ctest a; int main(int argc, char* argv[]) { boost::lexical_cast<string>(1); return 0; }

2 1

[Multi-Index] crash on invariant_() violation
by Dominique Devienne 02 Apr '09

02 Apr '09

We've recently introduced a boost multi_index_container into our code base. Now another seemingly unrelated change triggers an invariant violation in the bmi code during a unit test. I've tried to look at what kind of invariant is checked in the code, but got lost in its intricacies... Could I please get some pointers on what I should specifically look for, and maybe what invariant violations could be triggered from (or have been in the past)? Assertion failed: invariant_(), file c:\...\boost-1.38.0-debug\boost\multi_index_container.hpp, line 683 I don't have all the details of the crash yet, but the invariant is violated either during a call to empty() on the implicit index of the container (a unique_hashed index), or a call to erase() on that same index. Note that the indices use BOOST_MULTI_INDEX_CONST_MEM_FUN on a proxy object, which extracts the indexed value on the proxy-wrapped pointer. It's quite possible our code base makes a change to one of the indexed value without notifying the bmi. Is this something that would trigger an invariant violation? (it's on my todo list to avoid having something like this happening). Any insights in troubleshooting the above would be appreciated. Thanks, --DD

2 1

Re: [Boost-users] boost::array < pair<>... > iterator adaptor
by Hicham Mouline 02 Apr '09

02 Apr '09

>AMDG >Hicham Mouline wrote: >> I have a boost::array< pair<type1, type2>, maxsize > object, and I wish to >> iterate over the list of type1 elements, and the list of type2 elements... >> >> I gather I then need an iterator adaptor based on array's iterator and >> const_iterators... >> How may I write such a thing? >> >> I then can pass around iterators to type1 and type2 separately. >> >transform_iterator might be useful. I don't know off >the top of my head whether there is a more specialized >iterator adaptor in Boost somewhere. Great, this worked beautifully struct FirstElem : public std::unary_function< pair<>, double > { double operator()( const pair<>& p ) { return p.first; } }; typedef boost::transform_iterator< FirstElem, IterType > FirstElemIterType; FirstElemIterType feit ( container.begin() ); I can then pass feit around. Now, under normal assumptions, accessing first elements through feit should be as fast as direct access by manual loop.... inline calls should be optimized away? Thanks,

2 1

boost::array and operator< on private type (which is a pair) for using std algos
by Hicham Mouline 02 Apr '09

02 Apr '09

Hello, I'm not sure this is specific about boost::array... I apologize if it is not boost specific. I have this code http://codepad.org/h9uFoI1I This compiles under g++4.0 1. I find it strange that it does given that EntryType is a __private__ type inside C<T1,T2>. 2. It doesn't do what I expect: my operator< is defined in std namespace but is never called (i did this because with std::pair comes < operators frm the std namespace) I don't want to define operator< for all pairs, just for contexts (member functions of C) which use EntryType. Is this because I typedef EntryType? rds,

1 0