June 2008 - Boost-users - lists.preview.boost.org

xpressive: sregex_token_iterator assertion
by Jesse Williamson 24 Jun '08

24 Jun '08

Hi, folks! This program suprises me by causing a runtime assertion: ===== #include <string> #include <boost/xpressive/xpressive.hpp> using namespace boost::xpressive; int main() { std::string x("a\tb\tc"); sregex_token_iterator cur(x.begin(), x.end(), as_xpr("\t"), -1), end; for(; cur != end; ++cur) ; } ===== ...the assertion: ===== suprise: /usr/local/include/boost-1_35/boost/xpressive/detail/utility/tracking_ptr.hpp:196: void boost::xpressive::detail::enable_reference_tracking<Derived>::release() [with Derived = boost::xpressive::detail::regex_impl<__gnu_cxx::__normal_iterator<const char*, std::basic_string<char, std::char_traits<char>, std::allocator<char> > > >]: Assertion `0 < this->cnt_' failed. Aborted ===== Changing the declarations like this causes it to work as I expect: ===== sregex tab_delim(as_xpr("\t")); sregex_token_iterator cur(x.begin(), x.end(), tab_delim, -1), end; ===== ...maybe there's some "temporary" magic happening, or that I was just plain doing something I shouldn't have been doing (silly me!), or both? Could anyone comment on what's happening? [Note: Boost 1.35, OpenSuSE 10.3, g++ 4.1.2] Thanks! _Jesse Williamson ;-};

2 3

Re: [Boost-users] using transform_iterator to iterate through the values of a std::map
by Alan M. Carroll 24 Jun '08

24 Jun '08

At 11:06 AM 6/21/2008, you wrote: >Hi, > >This is probably a very newbie question. I'd like to iterate through the values of a std::map. I searched through the archives, and found this thread on this mailing list about it: Sorry for the double post, but I realized that Boost.Function can help with this . To avoid clutter, I have put my updated reply that more directly answers your question and tested example code on a weblog post: https://network-geographics.com/weblog/2008/06/boost_transform_iterator_use…

1 0

boost::asio connection timeout
by Ian O'Connell 24 Jun '08

24 Jun '08

Hello I was wonder if its possible to create synchronous behaviour regarding a connection attempt, i.e. Upper library layer calls : connect to host and waits for the connection to succeed or time out. asio can create the async connection attempt and timer and then run the ioservice to 'wait' on the async calls. which is fine in a simple senario where there are no other async calls running, ideally i'd like to be able to bring up a connection in a synchronous manner and then introduce the socket into an io_service which is processed by a pool of threads. Is it possible to either change the io_service a tcp socket uses, or retrieve the native type in a tcp socket and release the objects ownership of the type and assign this to a new object which uses the other io_service? i.e. Thread local io_service for setup library io_service which is processed by threadpool ? Thanks, Ian.

2 1

[BGL] AT&T Graphviz adaptor
by Dmitry 24 Jun '08

24 Jun '08

Hi there, I have a very nice BGL adaptor for AT&T graphviz graph representation. Is it worthy to add it to the boost library ? Thanks in advance, Sr. Dmitry

2 2

[DateTime] 1.35.0: Patch to fix typo
by peter.klotz＠ith-icoserve.com 24 Jun '08

24 Jun '08

The attached patch fixes a typo in boost/date_time/gregorian/greg_weekday.hpp. Best regards, Peter.

2 1

convert std::string to std::wstring
by Hansi 24 Jun '08

24 Jun '08

Hello, I have searched a lot in every place what is the best way to convert a string to a wstring (and may be vice versa). But til now I don`t got any results. Can anyone give me an hint what is the best way to do that (also good performance?)? maybe there is something available to do that? may be also with some examples. Thanks Hansjörg

6 8

[interprocess] example for boost::interprocess::map where the key is something in shared memory
by Jason Sachs 24 Jun '08

24 Jun '08

does anyone have a simple example of a map in shared memory, where the key is also in shared memory & not a simple "int"? I am trying to use a key of type boost::interprocess::basic_string<wchar_t, ...>, value of type unsigned long, and I am not sure about the subtleties of using boost::interprocess::map<>.insert() where the key lives in shared memory. (the example http://www.boost.org/doc/libs/1_35_0/doc/html/interprocess/quick_guide.html… is useful but both the key and value are simple types.) If I understand right, you have to insert() an instance of the std::pair<> class. But if the map is in shared memory, does this mean the std::pair<> also has to live in shared memory as well? (vs. a local std::pair<> that exists on the stack) For some reason the basic_string<wchar_t,...> seems to have been instantiated in some other weird area of memory instead & it's very difficult to debug + figure out what's going wrong.

2 4

[serialization] assert loading shared_ptr to abstract base class using boost 1.35.0
by Ult Mundane 24 Jun '08

24 Jun '08

Hi there, Using the boost::serialization library, I'm hitting an assert while deserializing. It happens only in a specific set of circumstances requiring a shared_ptr to an abstract base class. Class "Derived" is derived from class "AbstractBase" which is abstract. I'm serializing two objects which are of type shared_ptr<AbstractBase> that point to objects of type Derived. After the two shared_ptr's are serialized to disk, when I read them back in I hit an assert in basic_iarchive.cpp: "Assertion `new_cid == cid' failed". This assert happens on the second object only. As far as I can tell from the serialization documentation, I am using the appropriate BOOST_CLASS_EXPORT() line for the derived class, as well as registering the base-derived relationship with base_object(), and I even threw in a couple of BOOST_SERIALIZATION_SHARED_PTR() calls for good measure although I don't think they do anything in my environment. Two additional pieces of information: - If the base class is not abstract, but I still use BOOST_IS_ABSTRACT(), I get the same assert. This might indicate that some necessary polymorphism code isn't being generated as a result of declaring the class abstract. - If I serialize the pointers directly instead of using shared_ptr, it does not hit the assert. I have a workaround for this problem: it works as expected if I define the load_construct_data() template function for the base class do to nothing (assert), instead of using the BOOST_IS_ABSTRACT() macro. Attached is the minimal code to reproduce the problem. I'm using boost 1.35.0 with gcc 3.3.5 on Linux (RHEL4). Is there something I'm missing to make this work? Thanks in advance, Ult #include <fstream> #include <boost/archive/text_oarchive.hpp> #include <boost/archive/text_iarchive.hpp> #include <boost/serialization/shared_ptr.hpp> #include <boost/serialization/is_abstract.hpp> #include <boost/serialization/export.hpp> class AbstractBase { public: AbstractBase() : mInt(-1) { } virtual ~AbstractBase() { } virtual void func() const = 0; template<class Archive> void serialize(Archive & ar, const unsigned int) { ar & mInt; } int mInt; }; BOOST_IS_ABSTRACT(AbstractBase); BOOST_SERIALIZATION_SHARED_PTR(AbstractBase); BOOST_CLASS_IMPLEMENTATION(AbstractBase, boost::serialization::object_serializable); class Derived : public AbstractBase { public: Derived() : mInt2(-2) { } void func() const { } template<class Archive> void serialize(Archive & ar, const unsigned int) { ar & boost::serialization::base_object<AbstractBase>(*this); ar & mInt2; } int mInt2; }; BOOST_CLASS_EXPORT(Derived); BOOST_SERIALIZATION_SHARED_PTR(Derived); BOOST_CLASS_IMPLEMENTATION(Derived, boost::serialization::object_serializable); int main(int /* argc */, char * /* argv */[]) { char const * fname = "abstract.out"; boost::shared_ptr<AbstractBase> baseA(new Derived()); boost::shared_ptr<AbstractBase> baseB(new Derived()); { std::ofstream ofs(fname, std::ios::out | std::ios::binary); boost::archive::text_oarchive oa(ofs); oa & baseA; oa & baseB; } { std::ifstream ifs(fname, std::ios::in | std::ios::binary); boost::archive::text_iarchive ia(ifs); ia & baseA; std::cout << "The following line asserts:" << std::endl; ia & baseB; // Assertion `new_cid == cid' failed // Full text of assert: // // /dev/libs/boost_1_35_0/libs/serialization/src/basic_iarchive.cpp:432: const boost::archive::detail::basic_pointer_iserializer* boost::archive::detail::basic_iarchive_impl::load_pointer(boost::archive::detail::basic_iarchive&, void*&, const boost::archive::detail::basic_pointer_iserializer*, const boost::archive::detail::basic_pointer_iserializer*(*)(const boost::serialization::extended_type_info&)): Assertion `new_cid == cid' failed. } }

2 1

using transform_iterator to iterate through the values of a std::map
by Ákos Maróy 24 Jun '08

24 Jun '08

Hi, This is probably a very newbie question. I'd like to iterate through the values of a std::map. I searched through the archives, and found this thread on this mailing list about it: http://lists.boost.org/boost-users/2007/11/32374.php Unfortunately I can't put the partial recommendations together from that thread to work for me. While this is most probably due to my lack of experience with boost, can someone point to a sort-of complete usage solution to this question? Would be really appreciated. Akos

4 4

Calculating needed space for Boost.Serialization
by Ioannis Papadopoulos 24 Jun '08

24 Jun '08

Hi, I want to calculate exactly how much space it is needed to serialize a number of objects. What I'm thinking is another class that implements the Saving Archive concept and only invokes sizeof() rather than actually writing anything. From what I understand I should overload the << and & operators for this class to call save() or serialize() on the object, unless it is a POD (in which case the size is already known). Has this been done before? Thanks

2 2