- Boost-users - lists.boost.org

[lexical_cast] Canonical representation for numbers?
by Francesco Biscani 21 Sep '10

21 Sep '10

Hello, does lexical_cast make some kind of guarantees on the output of arithmetic types cast as strings? I'm asking because I need to pass a lexical representation of integers and floats to the constructors of some 3rd party class that expects as input strings containing a certain representation of numbers. E.g., a float represented as string must be either in decimal or scientific notation, with "." as separator, "e" or "E" preceding the exponent, leading sign only in case of negative numbers and so on. So far I've used successfully lexical_cast for such purposes, but I'm wondering if other factors (e.g., locale) could alter the string representation of numbers. Is it safe to assume all numbers will always be cast in the same string structure regardless of platform, i18n, etc.? Cheers, Francesco.

3 3

has_trivial_copy for optimization
by Hite, Christopher 21 Sep '10

21 Sep '10

I noticed poor performance coping a structure with a bunch of optionals. The reason is the cost of checking if the optionals are set. CPU's hate branching it breaks their pipelining. if(that->a) this->a=*that->a; At the end of the day I just wrote a copy construct which does a memcpy(), since the data types were just POD. It makes me a bit nervous since my code might break with a different impl of opitonal. Then I thought hey wouldn't it be cool if defined a metafunction that specified memcpy/memmove are allowed? It looks like has_trivial_copy<> fits the bill. // Suggested code: namespace boost{ template<typename T> struct has_trivial_copy< boost::optional<T> > : has_trivial_copy<T> {}; }//namespace boost typedef boost::optional<int> OptionalInt; BOOST_STATIC_ASSERT(( has_trivial_copy<OptionalInt>::value )); //fixed thanks to code above struct SomeOptionals{ OptionalInt a,b;}; BOOST_STATIC_ASSERT(( has_trivial_copy<SomeOptionals>::value )); //fails The second test fails. It can't imagine how to make it work without some compiler magic. I wonder if a compiler could use metafunctions to know which optimizations were legal. Similar meta function specializations could be written for boost::variant and boost::fusion::map. I could imagine contains like std::vector using this to do memmove() for example. Chris

2 1

Using Boost.Spirit with Boost.Units
by Vitaly Budovski 21 Sep '10

21 Sep '10

Hi, I need to be able to construct expressions from a collection of basic units and prefixes. The expression is provided as a string and thus needs to be parsed first. I have been able to achieve without too much trouble. What I am having trouble with is constructing the appropriated expressions using the Boost.Units library. I'm still very new to these two boost libraries so any pointers on how to achieve what I'm after (is it possible?) would be very much appreciated. Please see my code below: ----------------------------------- #include <boost/spirit/include/phoenix.hpp> #include <boost/spirit/include/qi.hpp> #include <boost/units/systems/si.hpp> #include <boost/units/systems/si/prefixes.hpp> #include <boost/units/systems/cgs.hpp> #include <boost/units/systems/temperature/celsius.hpp> #include <boost/variant.hpp> #include <iostream> #include <string> using namespace boost::units; using namespace boost::units::si; // Most prefixes omitted for brevity. struct si_prefix_symbol_: boost::spirit::qi::symbols<char, int> { si_prefix_symbol_() { add ("k", 3) ("M", 6) ("G", 9) ("c", -2) ("m", -3) ; } } si_prefix_symbol; typedef boost::variant<quantity<cgs::mass>, quantity<length>, quantity<si::time>, quantity<absolute<si::temperature> >, quantity<absolute<celsius::temperature> >, double> variant_t; struct unit_symbol_: boost::spirit::qi::symbols<char, variant_t> { unit_symbol_() { add ("m", 1.0 * meter) ("g", 1.0 * cgs::gram) ("s", 1.0 * second) ("K", 1.0 * absolute<si::temperature>()) ("C", 1.0 * absolute<celsius::temperature>()) ; } } unit_symbol; template<class Iterator> struct unit_expression: boost::spirit::qi::grammar<Iterator, variant_t()> { unit_expression(): unit_expression::base_type(start) { using boost::spirit::int_; using boost::spirit::ascii::space; using boost::spirit::eps; using namespace boost::spirit::arg_names; start = eps[_val = 1.0] >> ( ( ( unit_symbol | si_prefix_symbol >> unit_symbol ) >> -( '^' >> int_ ) ) % space ) ; } boost::spirit::qi::rule<Iterator, variant_t()> start; }; int main() { std::string in("kg m^-2"); typedef std::string::const_iterator iterator_type; typedef unit_expression<iterator_type> unit_expression; iterator_type first = in.begin(); iterator_type last = in.end(); unit_expression unit_parser; variant_t result; bool r = parse(first, last, unit_parser, result); if(r && first == last) { std::cout << "success: " << result << std::endl; } else { std::cout << "parsing failed at: "; std::cout << std::string(first, last) << std::endl; } return 0; }

1 0

[multi_array] pointer type - does not compile
by Suresh Kumar 21 Sep '10

21 Sep '10

Hi The following code does not compile. #include <boost/multi_array.hpp> #include <iostream> int main () { typedef boost::multi_array<int, 3> array_type; typedef array_type::index index; array_type * A; A = new array_type(boost::extents[3][4][2]); array_type::extent_gen extents; A[0][0][0] =1; return 0; } gives large number of errro messages of which some are given below. /usr/include/boost/multi_array/concept_checks.hpp:92: error: ‘int’ is not a class, struct, or union type /usr/include/boost/multi_array/concept_checks.hpp:71: error: using invalid field ‘boost::detail::multi_array::ConstMultiArrayConcept<Array, NumDims>::id’ I am using a pointer type because, the size of the multi_array is only known at run time and I want the multi_array to be a private data element of a class so that the methods can store data in my multi_array. Thanks for suggestions suresh -- R Suresh Kumar, http://www.ee.ucr.edu/~sramachandranna Sarve bhavantu sukinaha (May everyone be happy) Sarve santu niramiyaha (May everyone be without affliction) Sarve bhadrani pashyantu (May everyone see only goodness) Ma kashchit dukha bhak bhavet (Let none be the victim of suffering)

2 1

[asio]Why std::ofstream AND boost::asio::async_read() fails?
by Yasuhiko Yoshimura (informax.co.jp) 21 Sep '10

21 Sep '10

Hi All, I'm not able to see why downloading is sometimes fails in asio asyncread(). And I appriciate any suggestion. When 'comment===>> //' and 'insert ===>>' lines are valid, download file size is sometimes short than the original size from internet server(with big files). But if I put 'comment===>> //' back to no comment and remove 'insert ===>>' lines, the downloading result is perfect. class ftp_client_dtp { ... void start() { work_resolver_.reset(new boost::asio::ip::tcp::resolver(work_io_service_)); work_socket_.reset(new boost::asio::ip::tcp::socket(work_io_service_)); work_.reset(new boost::asio::io_service::work(work_io_service_)); work_thread_.reset(new boost::thread( boost::bind(&boost::asio::io_service::run, &work_io_service_))); ... comment===>> // std::string spath = scurrentpath_ + "\\" + sfilename_; comment===>> // f.open(spath.c_str(), std::ios::binary | std::ios_base::out); ... } ... void handle_read_content(const boost::system::error_code& err) { if (!err) { insert ===>> std::string spath = scurrentpath_ + "\\" + sfilename_; insert ===>> f.open(spath.c_str(), std::ios::binary | std::ios_base::out | std::ios_base::app); if (response_.size() > 0) { f << &response_; } insert ===>> f.close(); // Continue reading remaining data until EOF. boost::asio::async_read(*work_socket_, response_, boost::asio::transfer_at_least(1), boost::bind(&ftp_client_dtp::handle_read_content, shared_from_this(), boost::asio::placeholders::error)); } else if (err == boost::asio::error::eof) { // eof std::cout << err; insert ===>> std::string spath = scurrentpath_ + "\\" + sfilename_; insert ===>> f.open(spath.c_str(), std::ios::binary | std::ios_base::out | std::ios_base::app); if (response_.size() > 0) { f << &response_; } insert ===>> f.close(); do_close(); } else { std::cout << err; do_close(); } } void do_close() { comment===>> // f.close(); ... } std::ofstream f; std::string sip_, sport_, scurrentpath_, sfilename_; ... boost::asio::streambuf response_; ... };

1 0

[Serialization] Problem serializing directly and through base pointer in the same program.
by Travis Abbott 21 Sep '10

21 Sep '10

Hi, I am using Boost.Serialization to serialize classes through a shared_ptr to an abstract base class. Sometimes, I also serialize these classes directly. This was working for me in version 1.40, but not versions 1.41 or 1.44. If I serialize only shared_ptr<BaseClass> instances, everything is fine. If I link in code that serializes the derived class directly, even if that code is never called, serialization of shared_ptr<BaseClass> instances throws an exception ("unregistered class"). A few notes about what I am doing: I am calling BOOST_CLASS_EXPORT in the implementation file for each serializable class. I am using a template function for serialization that accepts any serializable type, creates a new text_oarchiver on the stack, and returns a string. This is called from many places in my application with many different types of things (making it hard to have a central place to put all the BOOST_CLASS_EXPORTs). I am having this problem on RHEL5 with gcc 4.1.2 WITHOUT optimization turned on. It works as expected with -O1 or higher. On WinXP with Visual Studio 2005, it fails in both release and debug mode. The problem also seems to depend on the order that the object files get written into the executable. If I include the file that is calling BOOST_CLASS_EXPORT first, things work as expected, but fail otherwise. (e.g., g++ obj.cpp main.cpp ... works, g++ main.cpp obj.cpp ... fails). Here are 3 source files, obj.h, obj.cpp, main.cpp that demonstrate this problem. The line in main: fromString(toString(obj), obj2); causes the code just before it (using shared_ptr) to fail. If anyone can let me know if I am misusing the library somehow, or if there is a cleaner way to accomplish what I am trying to do, I would be very appreciative. Thanks, Travis Abbott --------------------------8< obj.h 8< -------------------------- #pragma once #include <boost/serialization/void_cast.hpp> #include <string> class Base { public: virtual ~Base() {} virtual std::string name() const = 0; }; class Obj : public Base { public: std::string name() const { return "Obj"; } template<typename Arch> void serialize(Arch& arch, const unsigned version) { boost::serialization::void_cast_register<Obj, Base>( static_cast<Obj*>(NULL), static_cast<Base*>(NULL)); arch & data; } int data; }; --------------------------8< obj.cpp 8< -------------------------- #include "obj.h" #include <boost/archive/text_oarchive.hpp> #include <boost/archive/text_iarchive.hpp> #include <boost/serialization/export.hpp> BOOST_CLASS_EXPORT(Obj) --------------------------8< main.cpp 8< -------------------------- #include "obj.h" #include <boost/serialization/shared_ptr.hpp> #include <boost/archive/text_iarchive.hpp> #include <boost/archive/text_oarchive.hpp> #include <boost/shared_ptr.hpp> #include <iostream> #include <sstream> #include <stdexcept> #include <string> using boost::shared_ptr; using namespace std; template<typename T> string toString(const T& obj) { ostringstream ss; boost::archive::text_oarchive arch(ss); arch << obj; return ss.str(); } template<typename T> void fromString(const string& s, T& obj) { istringstream ss(s); boost::archive::text_iarchive arch(ss); arch >> obj; } int main() { try { shared_ptr<Base> pobj(new Obj), pobj2; fromString(toString(pobj), pobj2); cout << pobj2->name() << endl; // Compiling this causes runtime failure in code above. Obj obj, obj2; fromString(toString(obj), obj2); cout << obj2.name() << endl; } catch (const exception& e) { cerr << "Error: " << e.what() << endl; } return 0; }

2 2

boost serialisation 1.44 /linux /gcc 4.2.1 boost serialisation 1.44 /linux /gcc 4.2.1 boost serialisation 1.44 /linux /gcc 4.2.1 boost serialisation 1.44 /linux /gcc 4.2.1 boost serialisation 1.44 /linux /gcc 4.2.1 Re:boost serialisatio
by Avi Bahra 21 Sep '10

21 Sep '10

> Having problems with boost 1.44 and serialisation on Linux, gcc 4.2.1 > I am migrating from boost 1.43 to boost 1.44 > and get compiler errors of the form: > > In file included from ACore/src/TimeSeries.hpp:13, > from ACore/test/TestTimeSeries.cpp:16: > /var/tmp/ma0/boost/boost_1_44_0/boost/date_time/posix_time/time_serialize.hpp: > In function ?void boost::serialization::save(Archive&, const > boost::posix_time::time_duration&, unsigned int)?: > /var/tmp/ma0/boost/boost_1_44_0/boost/date_time/posix_time/time_serialize.hpp:42: > error: there are no arguments to ?make_nvp? that depend on a template > parameter, so a declaration of ?make_nvp? must be available > > This is surprising as boost 1.44 release notes makes no mention of any > changes to boost serialisation ? > > Can anyone help ? > Bryce Lelbach wrote: >>Looks like a problem with date_time, not serialization. I know 1.44 builds on >>GNU/Linux GCC 4.2.1 for me. >> >>Can you make sure you have a file at >>/var/tmp/ma0/boost/boost_1_44_0/boost/serialization/nvp.hpp? If you do have it, >>could you show us? I just downloaded boost 1.44 today: The file: /var/tmp/ma0/boost/boost_1_44_0/boost/serialization/nvp.hpp exist. This is what I have: //=============================================================== #ifndef BOOST_SERIALIZATION_NVP_HPP #define BOOST_SERIALIZATION_NVP_HPP // MS compatible compilers support #pragma once #if defined(_MSC_VER) && (_MSC_VER >= 1020) # pragma once #endif /////////1/////////2/////////3/////////4/////////5/////////6/////////7/////////8 // nvp.hpp: interface for serialization system. // (C) Copyright 2002 Robert Ramey - http://www.rrsd.com . // Use, modification and distribution is subject to the Boost Software // License, Version 1.0. (See accompanying file LICENSE_1_0.txt or copy at // http://www.boost.org/LICENSE_1_0.txt) // See http://www.boost.org for updates, documentation, and revision history. #include <utility> #include <boost/config.hpp> #include <boost/detail/workaround.hpp> // supress noise #if BOOST_WORKAROUND(BOOST_MSVC, <= 1200) # pragma warning (disable : 4786) // too long name, harmless warning #endif #include <boost/mpl/integral_c.hpp> #include <boost/mpl/integral_c_tag.hpp> #include <boost/serialization/level.hpp> #include <boost/serialization/tracking.hpp> #include <boost/serialization/split_member.hpp> #include <boost/serialization/base_object.hpp> #include <boost/serialization/traits.hpp> #include <boost/serialization/wrapper.hpp> namespace boost { namespace serialization { template<class T> struct nvp : public std::pair<const char *, T *>, public wrapper_traits<const nvp<T> > { explicit nvp(const char * name_, T & t) : // note: redundant cast works around borland issue // note: added _ to suppress useless gcc warning std::pair<const char *, T *>(name_, (T*)(& t)) {} nvp(const nvp & rhs) : // note: redundant cast works around borland issue std::pair<const char *, T *>(rhs.first, (T*)rhs.second) {} const char * name() const { return this->first; } T & value() const { return *(this->second); } const T & const_value() const { return *(this->second); } // True64 compiler complains with a warning about the use of // the name "Archive" hiding some higher level usage. I'm sure this // is an error but I want to accomodated as it generates a long warning // listing and might be related to a lot of test failures. // default treatment for name-value pairs. The name is // just discarded and only the value is serialized. template<class Archivex> void save( Archivex & ar, const unsigned int /* file_version */ ) const { // CodeWarrior 8.x can't seem to resolve the << op for a rhs of "const T *" ar.operator<<(const_value()); } template<class Archivex> void load( Archivex & ar, const unsigned int /* file_version */ ){ // CodeWarrior 8.x can't seem to resolve the >> op for a rhs of "const T *" ar.operator>>(value()); } BOOST_SERIALIZATION_SPLIT_MEMBER() }; template<class T> inline #ifndef BOOST_NO_FUNCTION_TEMPLATE_ORDERING const #endif nvp<T> make_nvp(const char * name, T & t){ return nvp<T>(name, t); } // to maintain efficiency and portability, we want to assign // specific serialization traits to all instances of this wrappers. // we can't strait forward method below as it depends upon // Partial Template Specialization and doing so would mean that wrappers // wouldn't be treated the same on different platforms. This would // break archive portability. Leave this here as reminder not to use it !!! #if 0 // #ifndef BOOST_NO_TEMPLATE_PARTIAL_SPECIALIZATION template <class T> struct implementation_level<nvp<T> > { typedef mpl::integral_c_tag tag; typedef mpl::int_<object_serializable> type; BOOST_STATIC_CONSTANT(int, value = implementation_level::type::value); }; // nvp objects are generally created on the stack and are never tracked template<class T> struct tracking_level<nvp<T> > { typedef mpl::integral_c_tag tag; typedef mpl::int_<track_never> type; BOOST_STATIC_CONSTANT(int, value = tracking_level::type::value); }; #endif } // seralization } // boost #include <boost/preprocessor/stringize.hpp> #define BOOST_SERIALIZATION_NVP(name) \ boost::serialization::make_nvp(BOOST_PP_STRINGIZE(name), name) /**/ #define BOOST_SERIALIZATION_BASE_OBJECT_NVP(name) \ boost::serialization::make_nvp( \ BOOST_PP_STRINGIZE(name), \ boost::serialization::base_object<name >(*this) \ ) /**/ #endif // BOOST_SERIALIZATION_NVP_HPP // ======================================== Best regards, Ta, Avi

2 1

Using boost::interprocess shared memory for both 32-bit and 64-bit processes
by Jonathon 21 Sep '10

21 Sep '10

Hello, I would like to read/write from/to boost::interprocess shared memory segments from a 32-bit process and a 64-bit process on the same machine. Is this possible? I've attempted to do this, but I seem to be experiencing some compatibility issues. Note that I am reading/writing boost::inteprocess containers to shared memory. For example, in my 32-bit process, I'll add some data to a boost::interprocess::vector and when I read this data from my 64-bit process, it returns garbage. Are there compiler options I can set to make things compatible? If it is not possible to do this, what are some possible solutions for this? Thanks, J

3 2

[multi_array] why this does not compile? allocator?
by Suresh Kumar 20 Sep '10

20 Sep '10

Could you kindly tell me why the following code does not compile? I guess it is an allocator issue, but how do I go further? #include <boost/multi_array.hpp> #include <iostream> using namespace std; int main () { class test{ int x; int y; }; typedef boost::multi_array<test, 3> array_type; typedef array_type::index index; array_type A(boost::extents[3][4][2]); array_type::extent_gen extents; return 0; } compile errors: multiArray.cpp:11: error: template argument for ‘template<class _Alloc> class std::allocator’ uses local type ‘main()::test’ multiArray.cpp:11: error: trying to instantiate ‘template<class _Alloc> class std::allocator’ multiArray.cpp:11: error: template argument 3 is invalid multiArray.cpp:11: error: invalid type in declaration before ‘;’ token multiArray.cpp:12: error: expected initializer before ‘index’ multiArray.cpp:13: error: cannot convert ‘boost::detail::multi_array::extent_gen<3ul>’ to ‘array_type’ in initialization multiArray.cpp:14: error: expected initializer before ‘extents’ -- R Suresh Kumar, http://www.ee.ucr.edu/~sramachandranna Sarve bhavantu sukinaha (May everyone be happy) Sarve santu niramiyaha (May everyone be without affliction) Sarve bhadrani pashyantu (May everyone see only goodness) Ma kashchit dukha bhak bhavet (Let none be the victim of suffering)

2 1

[Fusion] fusion::for_each with stateful functors
by Paul Graphov 20 Sep '10

20 Sep '10

Hello Boost Users, More than three years ago this question arose in this list: http://lists.boost.org/boost-users/2007/03/26338.php And as I understand it was concluded that overload for non-const references should be added. But now there are only const references to functor available. Was it forgotten or there are some reasons not to have such overload? What is the best way to solve the problem? Mutable members or some proxies look quirky. Thanks!

3 3