- Boost-users - lists.boost.org

[Serialization] Postpone saving
by Mailing Lists Controller 18 Jul '10

18 Jul '10

Hi, I'm working with the Serialization library and in detail with the xml archives. In my code I have to serialize a lot of classes and I wish to save in the xml file only when I've done with the serialization of all clasess, also I would like to have more control when the Serialization saves the file (do it raise exceptions when it save files? Which ones?). As far as I read from the tutorials when I use the operator<< in the xml_archive it save directly to the files. In my test I've seen that the archive have no problem to save many classes to the same archive however as I've wrote I would like to have more control (at least to know which errors appear during writing). Is it possible? Thank you for your time.

3 2

How to fwd decl boost::variant (with type sequences)?
by Paul 17 Jul '10

17 Jul '10

In a large project (winxp, boost 1.38.0, 10K sources) we are using the boost::variant at some places. The variant types are visible to a large amount of cpp’s. Many of these source-file are not ‘using’ the variant at all, a small amount only ‘passes’ a variant type, and an even smaller amount actually uses it (copy, static-visitor, etc). We have noticed however that the compile-time increases for all sources-files, even when the variant is not used. On a single source-file its maybe split of a second extra but with 10K sources it quickly becomes problem. First attempt to solve this is to fwd declare the variant using provided header variant_fwd.hpp. Example: [v.h] #include <boost/variant/variant_fwd.hpp> class I001{ public: }; class I002{ public: }; class I003{ public: }; typedef boost::variant< boost::shared_ptr<C001>, boost::shared_ptr<C002>, boost::shared_ptr<C003> > v; [a.h] #include “v.h” class CA { public: void foo(const v& ref); }; Source-files now compile faster while only the sources that actually use the variant need to add #include <boost/variant/variant.hpp> to get the definition of the variant class. However it doesn’t really fit our needs because we need the ‘counted’ variant types (using typelists). Reasons for this is simply that some variants are only 3, 5 or 10 elements while another holds 39 element types. We don’t want to extend the length of each type to say 50. Therefore we first compose an mpl::vector (and even compose subsets using joint_view) before defining the variant types. Example of use: #include <boost/mpl/vector.hpp> #include <boost/variant.hpp> typedef boost::mpl::vector< boost::shared_ptr<I001>, boost::shared_ptr<I002>, boost::shared_ptr<I003> > MplV; typedef boost::make_variant_over<MplV>::type v; Now each cpp file must not only include the variant definitions again but also (part of) mpl. How could we solve this issue? One idea we had is to wrap the variant in a class and fwd decl that. Although it works I’m not sure if that is the way to go? Thanks for any advise, Paul

1 0

[phoenix][units] make boost.phoenix play well with boost.units
by alfC 16 Jul '10

16 Jul '10

I have and expression of type arg1*arg1 which is contained in a deduced type LambdaExp (more preciselly composite<multiplies_eval,vector<argument<0>, argument<0> >) when trying to see what is the phoenix-deduced return type when input is double typename LambdaExp::template result<basic_environment<double> >::type naturally, it gives 'double' (double*double) but when the input type is a boost::units::quantity typename LambdaExp::template result<basic_environment<quantity<si::length> > >::type it gives: error_cant_deduce_type (instread of the equivalent of quantity<si::area>, like in quantity<si::area> a = quantity<si::length>()*quantity<si::length>() ) this is strange since it seemed that Phoenix worked out of the box with Units. What is going on? How can we help Phoenix deduce the return type? Does Phoenix need this help, like Boost.Lambda does by means of boost/ units/lambda.hpp? Thank you, Alfredo

5 14

[fusion]push_back<list<>,int> > fails compilation
by Larry Evans 16 Jul '10

16 Jul '10

The code in the 1st attachment fails to compile as shown in 2nd attachment. If that's intended, the what's the justification for not allowing appending to an empty sequence? If that's not intended, then should I file a bug report? Based on: <fusion.doc>/html/fusion/algorithm/transformation/functions/push_back.html it seems push_back should work on a 'Forward Sequence' and based on: <fusion.doc>/html/fusion/container/list.html list is a 'Forward Sequence'. And based on the following quote from the Synopsis of the list.html: > The variadic class interface accepts 0 to FUSION_MAX_LIST_SIZE elements an empty list type, i.e. list<>, is a list, the attached code should compile. -regards Larry ###-*- mode: compilation; default-directory: "~/prog_dev/boost-svn/ro/trunk/libs/fusion/sandbox/" -*- Compilation started at Fri Jul 16 02:08:38 make HOW=gcc4_5n run install -d `dirname /home/evansl/prog_dev/boost-svn/ro/trunk/sandbox-local/build/gcc4_5n/boost-svn/ro/trunk/libs/fusion/sandbox/list_push_back.o` /home/evansl/download/gcc/4.5-20100624/install/bin/g++ -c -Wall -ftemplate-depth-300 -O0 -g3 -fno-inline -std=gnu++0x -I/home/evansl/prog_dev/boost-svn/ro/trunk -DTEMPLATE_DEPTH=300 list_push_back.cpp -MMD -o /home/evansl/prog_dev/boost-svn/ro/trunk/sandbox-local/build/gcc4_5n/boost-svn/ro/trunk/libs/fusion/sandbox/list_push_back.o list_push_back.cpp: In function 'void boost::fusion::sandbox::run()': list_push_back.cpp:25:45: error: no matching function for call to 'boost::fusion::joint_view<boost::fusion::list<>, const boost::fusion::single_view<int> >::joint_view(boost::fusion::result_of::push_back<const boost::fusion::list<>, int>::type)' /home/evansl/prog_dev/boost-svn/ro/trunk/boost/fusion/view/joint_view/joint_view.hpp:55:9: note: candidates are: boost::fusion::joint_view<Sequence1, Sequence2>::joint_view(Sequence1&, Sequence2&) [with Sequence1 = boost::fusion::list<>, Sequence2 = const boost::fusion::single_view<int>] /home/evansl/prog_dev/boost-svn/ro/trunk/boost/fusion/view/joint_view/joint_view.hpp:34:5: note: boost::fusion::joint_view<boost::fusion::list<>, const boost::fusion::single_view<int> >::joint_view(const boost::fusion::joint_view<boost::fusion::list<>, const boost::fusion::single_view<int> >&) make: *** [/home/evansl/prog_dev/boost-svn/ro/trunk/sandbox-local/build/gcc4_5n/boost-svn/ro/trunk/libs/fusion/sandbox/list_push_back.o] Error 1 Compilation exited abnormally with code 2 at Fri Jul 16 02:08:40

2 1

Re: [Boost-users] Annoying compiler warnings in asio/error.hpp
by Hite, Christopher 16 Jul '10

16 Jul '10

> What you describe here is the opposite of what is stated in the lib. > As far as I understand the author's intention by declaring a global variable > static: introdcution of an own copy of this var in each compilation unit. The author just wanted a constant which you ask for a name/message. Actually the variable ssl_category is not used anywhere. They call get_ssl_category() which has a static function variable. That is another good trick to create a global variable which exists in all compilation scopes. Code like this in a header will generate on instance of X shared between compilation scopes. Class X{} inline X& getOneAndOnlyX(){ static X x; return x; } Anyways I'm guessing you could just kick ssl_category. I can't find these constants documented anywhere in the reference. ssl_category isn't used in tests. Some of the others are used, but since they are the same as what the inline functions would return. Perhaps the author was afraid in a multi threaded environment functions like getOneAndOnlyX() aren't safe. I know gcc by default generates atomic ops so only one X is constructed. Perhaps the idea was to ensure all these things were created before main() gets a chance to start threads. I don't have a checkout. Maybe someone could svn blame the file and find out who defined these things and ask them what they were thinking. Chris

1 0

[Serialization] Why not allowing reset_object_address on object serialized throw a pointer ?
by Alain KOCELNIAK 16 Jul '10

16 Jul '10

This issue has already been discussed but was never addressed : http://lists.boost.org/boost-users/2008/11/42417.php https://svn.boost.org/trac/boost/ticket/4044 I don't understand why objects loaded as pointer can't be relocated : - These objects are stored in basic_iarchive_impl::object_id_vector as the others, so, technically speaking, we can change their "address" - There is situations where it is mandatory : for example when data, sharing some objects, are serialized / dispatched in separate archives => without relocating service at loading, sharing of object is not possible. For the moment, I workaround the limitation with a hack in the method basic_iarchive_impl::reset_object_address : I remove the "safeguard" which checks that the method is applied on a tracked object and immediately after the concerned object is read : ================================================ for(i = moveable_objects_recent; i < moveable_objects_end; ++i){ if(old_address == object_id_vector[i].address) break; } for(; i < moveable_objects_end; ++i){ ================================================ is replaced by ================================================= for(i = 0; i < object_id_vector.size(); ++i){ ================================================= I think that the limit can be properly removed by handling in "basic_iarchive_impl::load_pointer" the "moveable_objects_..." variables as it is done in "basic_iarchive_impl::load_object" As objects created by the serialization system are marked with "loaded_as_pointer" = true (this data is used by basic_iarchive_impl::delete_created_pointers) I think this data should be set to false by "reset_object_address". Is there other implication ? Alain.

1 0

returning pointer to Boost::assign::map_list_of
by alapex0310＠gmail.com 16 Jul '10

16 Jul '10

Hi, I am new to Boost::Assign. I have a std::map<std::string, unsigned char> and I have a SomeClass with constructor SomeClass(std::string, std::map<std::string, unsigned char>*) and I want to pass the pointer to the result of map_list_of to SomeClass like this: SomeClass *psc = new SomeClass("SomeClassName",map_list_of("one",1)("two",2)("three",3)("four",4)("five",5); In the example at http://www.boost.org/doc/libs/1_43_0/libs/assign/doc/index.html#intro , the map_list_of is definitely returning a small map. I don't want to pass the map object itself but the pointer to the map object returned by map_list_of. Please help me. -Best regards

2 1

[asio] creating acceptor on the stack causes seg-fault
by Steve Lorimer 16 Jul '10

16 Jul '10

Hi folks, I have the following class: using namespace boost::asio; // basic tcp server class server { boost::thread thread; ip::tcp::acceptor acceptor; io_service io_svc; // tcp_srv_conn object which will become the local end-point of the next incoming connection tcp_srv_conn_p new_conn; void loop(unsigned short port); void on_accept(const boost::system::error_code &err); public: server(); virtual ~server(); void serve(unsigned short port); void serve(const std::string &svc); void close(); }; and the constructor for server looks like this: server::server() : acceptor(io_svc), new_conn(new tcp_srv_conn(io_svc)) { } If I create a server object on the stack: server s; I get a segmentation fault deep inside the asio internals (something to do with a mutex), originating from initialising acceptor with my io_service object: acceptor(io_svc) Here is a stack trace of the core file: Marked in red is (#6) my server constructor and (#7) the basic_socket_acceptor constructor Core was generated by `tcp_srv.debug -p 12'. Program terminated with signal 11, Segmentation fault. #0 0x00007f8de6c3ed21 in pthread_mutex_lock () from /lib64/libpthread.so.0 (gdb) backtrace #0 0x00007f8de6c3ed21 in pthread_mutex_lock () from /lib64/libpthread.so.0 #1 0x0000000000421153 in boost::asio::detail::posix_mutex::lock (this=0x8) at /usr/include/boost/asio/detail/posix_mutex.hpp:71 #2 0x0000000000422db7 in scoped_lock (this=0x7f8de58d7a70, m=...) at /usr/include/boost/asio/detail/scoped_lock.hpp:36 #3 0x0000000000426e22 in boost::asio::detail::service_registry::use_service<boost::asio::socket_acceptor_service<boost::asio::ip::tcp> > (this=0x0) at /usr/include/boost/asio/detail/service_registry.hpp:94 #4 0x000000000042573a in boost::asio::use_service<boost::asio::socket_acceptor_service<boost::asio::ip::tcp> > (ios=...) at /usr/include/boost/asio/impl/io_service.ipp:195 #5 0x000000000042427d in basic_io_object (this=0x7f8de58d7bd0, io_service=...) at /usr/include/boost/asio/basic_io_object.hpp:72 #6 0x0000000000422fd7 in basic_socket_acceptor (this=0x7f8de58d7bd0, io_service=...) at /usr/include/boost/asio/basic_socket_acceptor.hpp:77 #7 0x000000000041fe63 in server (this=0x7f8de58d7b90) at src/tcp_srv.cpp:9 #8 0x000000000040d4d2 in app::start_here (this=0x7fffaab13b2f, ac=3, av=0x7fffaab13cd8) at test/server_test.cpp:65 Am I doing something stupid? How should I be creating my io_service and acceptor objects? TIA Steve

1 1

boost::serialization trouble
by Sebastian Weber 16 Jul '10

16 Jul '10

Hi! I am experiencing quite odd errors with the serialization library: My program compiles fine, but crashes when static constants are declared, i.e. before the program even starts with main(). The debugger tells me that some singleton declared in the seralization library conflicts: Program received signal EXC_BAD_ACCESS, Could not access memory. Reason: KERN_INVALID_ADDRESS at address: 0x0000000000000000 0x0000000100003aaf in H5::DataType::DataType () at singleton.hpp:149 149 BOOST_DLLEXPORT T & singleton<T>::instance = singleton<T>::get_instance(); Any suggestions would be great here. What makes it even harder to understand is the fact that the exact same code runs perferctly on a Linux machine, whereas the crash is on MacOS 10.6. My setup here is a boost 1.43.0 compiled with gcc 4.5 from pacports (or apples gcc 4.2, no difference). It seems that the singleton does things which are not allowed on a 10.6 MacOS machine... Many thanks for help, Sebastian Weber

2 3

boost lambda wiht bind to stringstream::str
by Conoscenza Silente 16 Jul '10

16 Jul '10

Hi all I am trying to get familiar with boost lambda. Below I am using std::transform algorithm but when I try to call std::stringstream::str() function using lambda::bind I receive the following error message: TestBoost15Jul.cpp:49: error: no matching function for call to ‘bind(<unresolved overloaded function type>, const boost::reference_wrapper<std::basic_stringstream<char, std::char_traits<char>, std::allocator<char> > >)’ Do you also know if is the right way to return the string that must be stored in the second vector used by std::transform? Regards AFG // C++ #include <iostream> #include <algorithm> #include <vector> #include <sstream> #include <string> // boost #include <boost/assign.hpp> #include <boost/assign/list_of.hpp> #include <boost/assign/std/vector.hpp> #include <boost/lambda/core.hpp> #include <boost/lambda/lambda.hpp> #include <boost/lambda/bind.hpp> #include <boost/function.hpp> int main( int argc, char** argv ){ namespace bl = boost::lambda; using namespace boost::assign; std::vector< std::string > vIn = list_of("a")("b")("c")("d"); std::vector< std::string > vOut; std::stringstream outFmt; int nCount = 2; bl::var_type< int >::type _nCount( bl::var(nCount) ); bl::var_type< std::stringstream >::type _outFmt( bl::var(outFmt) ); std::transform( vIn.begin(), vIn.end(), vOut.begin(), ( _outFmt << bl::_1 << "=%" << _nCount << "%", _nCount++, bl::bind( &std::stringstream::str , boost::ref(outFmt) )() // this call to bind has some problem ) ); std::cout << outFmt.str() << std::endl; return 1; }

2 6