- Boost - lists.boost.org

[Serialization] serialize from and to an already instantiated object.
by nico 10 Jun '11

10 Jun '11

Hello. I've already posted this topic in the boost-user mailing list, but no reply: maybe this boost list is more active... ? Anyway: I use boost.serialization to perform save and load to files. If a class is serializable through the boost mechanism, I'd like to easily add the "savable" feature (through serialization and boost archives). Here how I do, and then I've got 2 questions. I don't post all the #include statements. Let's have a serializable class (I only use text archive for this example) ---------MyClass.h--------- class MyClass { friend class boost::serialization::access; template<class Archive> void serialize(Archive & ar, const unsigned int version); int data_; public: MyClass(); virtual ~MyClass(); }; BOOST_CLASS_EXPORT_KEY(MyClass) ---------MyClass.cpp--------- MyClass::MyClass() : data_(123) {} MyClass::~MyClass() {} template<class Archive> void MyClass::serialize(Archive & ar, const unsigned int version) { ar & data_; } // instanciate the serialize function for text archives #include "boost/archive/text_iarchive.hpp" #include "boost/archive/text_oarchive.hpp" template void MyClass::serialize(boost::archive::text_iarchive & ar, const unsigned int version); template void MyClass::serialize(boost::archive::text_oarchive & ar, const unsigned int version); BOOST_CLASS_EXPORT_IMPLEMENT(MyClass) --------- Now I define a FileSavable class. Its purpose is to add the "save-to-file" feature to a class that can be serialized in a text archive, making the latter inherting from FileSavable, and adding a line in its serialize function, I write it below). ---------FileSavable.h--------- class FileSavable { friend class boost::serialization::access; template<class Archive> void serialize(Archive & ar, const unsigned int version); public: virtual ~FileSavable() {} void SaveToFile(std::string path); void LoadFromFile(std::string path); }; BOOST_CLASS_EXPORT_KEY(FileSavable) ---------FileSavable.cpp--------- template<class Archive> void FileSavable::serialize(Archive & ar, const unsigned int version) {} void FileSavable::SaveToFile(std::string path) { std::ofstream ofs(path.c_str()); assert(ofs.good()); boost::archive::text_oarchive oa(ofs); oa & this; } void FileSavable::LoadFromFile(std::string path) { std::ifstream ifs(path.c_str()); assert(ifs.good()); boost::archive::text_iarchive ia(ifs); FileSavable* p; ia & p; *this = *p; delete p; } // instanciate the serialize function for text archives #include "boost/archive/text_iarchive.hpp" #include "boost/archive/text_oarchive.hpp" template void FileSavable::serialize(boost::archive::text_iarchive & ar, const unsigned int version); template void FileSavable::serialize(boost::archive::text_oarchive & ar, const unsigned int version); BOOST_CLASS_EXPORT_IMPLEMENT(FileSavable) ---------Modification to MyClass--------- It now inherit from FileSavable: class MyClass : public FileSavable The serialize function call the base class: template<class Archive> void MyClass::serialize(Archive & ar, const unsigned int version) { ar & boost::serialization::base_object<FileSavable>(*this); ar & data_; } --------- Thanks if you read me until here, now my questions: 1) I don't care the FileSavable part of MyClass being serialized, but if I don't call it with boost::serialization::base_object, I have a runtime error: terminate called after throwing an instance of 'boost::archive::archive_exception' what(): unregistered void cast 7MyClass<-11FileSavable Do you see a way to avoid having to add this line in MyClass? Ideally, just having to inheriting from FileSavable would be great: a simple way to add a feature. 2) As you can see, my LoadFromFile function first creates a new MyClass object in p, by de-serializing an archive, and then copies this newly created object into "*this", and finally deletes the object created by the archive d-eserialization. I didn't find a way to avoid this temporary and de-serialize directly in this. Moreover, my solution is not exception-safe, but I didn't manage to de-serialize in a temporary auto_ptr nor boost::shared_ptr. Thanks in advance for replies. Nicolas.

3 3

[multi_array]inner_product interest?
by Larry Evans 10 Jun '11

10 Jun '11

Attached is code to calculate the inner product of 2 multi_arrays. As indicated in comments, the code was adapted from the Budd book. It may be faster than doing subscripting to get the array elements; however, I've done no benchmarks, but I'd guess it would be faster since there's no calls to subscripting operators. Instead, there's just one div call and 3 multiplications per element of the result array. The .txt attachment shows relevant quotes from the book and reasons why changes were made to the book's code to accommodate multi_array's different storage orders. I've not done too much testing, but did do a simple test for 2 by 2 arrays whose output seems correct. Any interest? -regards, Larry

1 1

Re: [boost] Is Boost interested in the Boost-based Base64 library
by Denis Shevchenko 10 Jun '11

10 Jun '11

> Hmmm - so you are proposing that something like the following header > be added to boost? > > ... Yes, Robert, something like this. > a) where in the directory/namespace would this be? boost, boost/utility, or? For example, in boost/base64 directory and in boost namespace. I think it CAN be in boost namespace. So we have boost/base64 directory with docs, examples and base64.hpp file. In main boost directory will be file with same name base64.hpp with: #include <boost/base64/base64.hpp> > c) All this for one small special purpose function? Yes, Robert, for one small, concrete and frequent task. > d) How is a user going to find this pair of functions in the boost libraries. As well as he going to find boost::mem_fn, for example... :-) > e) As soon as someone puts this in, imediately some else will ask - great > but it needs a parameter to so I can change the line width. Then someone > asks for someother tweak. This is the result of making a "simple solution". > It's easy enough to make a simple solution - it's very hard to keep it simple. I understand you. My solution is simple and will be simple. Without "line width", special formatting, etc. Base64 encoding/decoding, nothing else. - Denis

1 0

Re: [boost] Is Boost interested in the Boost-based Base64 library
by Sean Chittenden 09 Jun '11

09 Jun '11

I'm assuming folks have looked at using: > http://code.google.com/p/stringencoders/ Pretty simple API: > char buf[100]; > char result[10]; > char endian[] = {(char)0, (char)0, (char)1}; > int d = modp_b64_encode(buf, endian, 3); And very fast: > http://code.google.com/p/stringencoders/#Why With some neat, very platform specific tricks: > http://code.google.com/p/stringencoders/#How_It_Works Don't get me wrong, I'd definitely like to see Boost have something because that extreme of performance isn't always necessary, but FYI for those interested in the discussion. -sc -- Sean Chittenden sean(a)chittenden.org

1 0

Re: [boost] [Boost-users] More boostcon videos
by Marshall Clow 09 Jun '11

09 Jun '11

On Jun 9, 2011, at 9:05 AM, Ovanes Markarian wrote: > On Thu, Jun 9, 2011 at 5:53 PM, Marshall Clow <mclow.lists(a)gmail.com> wrote: > Bartoz (Tuesday): > http://blip.tv/boostcon/haskell-the-pseudocode-language-for-c-template-meta… > > Tony (Wednesday): > http://blip.tv/boostcon/lockfree-programming-part-2-data-structures-5258064 > > Cliff (Wednesday); > http://blip.tv/boostcon/design-journeys-with-skoot-a-peer-oriented-generic-… > > > I'm hoping to get about half of the videos up before I go on holidays next week. > > Hello Marshall, > > seems like both videos > http://blip.tv/boostcon/lockfree-programming-part-2-data-structures-5258064 > http://blip.tv/boostcon/design-journeys-with-skoot-a-peer-oriented-generic-… > > contain Cliff's presentation. D'oh! I fat-fingered the upload. New link for Tony's preso: http://blip.tv/boostcon/lockfree-programming-part-2-data-structures-5258642 -- Marshall Marshall Clow Idio Software <mailto:mclow.lists@gmail.com> A.D. 1517: Martin Luther nails his 95 Theses to the church door and is promptly moderated down to (-1, Flamebait). -- Yu Suzuki

1 0

[Accumulators + Signals2] Compilation error with GCC when both are used
by sguazt 09 Jun '11

09 Jun '11

Hello, It seems that Accumulators and Signals2 have problems to coexist. For instance, look at the following simple C++ program: --- [c++] --- #include <boost/accumulators/accumulators.hpp> #include <boost/accumulators/statistics/stats.hpp> #include <boost/accumulators/statistics/mean.hpp> #include <boost/signals2.hpp> int main() { } --- [/c++] --- If I compile it with GCC 4.6.0 (under Fedora Linux 15 x86_64): $ g++ -Wall -Wextra -pedantic -ansi -I,/boost-trunk -o boost_signals2_accumulators boost_signals2_accumulators.cpp I get the following errors: --- [error] --- In file included from ./boost-trunk/boost/signals2.hpp:20:0, from boost_signals2_accumulators.cpp:4: ./boost-trunk/boost/signals2/signal_type.hpp:29:2: error: #error This header requires BOOST_PARAMETER_MAX_ARITY to be defined as 7 or greater prior to including Boost.Parameter headers In file included from ./boost-trunk/boost/signals2.hpp:20:0, from boost_signals2_accumulators.cpp:4: ./boost-trunk/boost/signals2/signal_type.hpp:78:9: error: wrong number of template arguments (6, should be 5) ./boost-trunk/boost/parameter/parameters.hpp:714:8: error: provided for ‘template<class PS0, class PS1, class PS2, class PS3, class PS4> struct boost::parameter::parameters’ ./boost-trunk/boost/signals2/signal_type.hpp:83:9: error: ‘parameter_spec’ is not a class or namespace ./boost-trunk/boost/signals2/signal_type.hpp:83:29: error: expected unqualified-id before ‘<’ token ./boost-trunk/boost/signals2/signal_type.hpp:99:11: error: ‘args’ was not declared in this scope ./boost-trunk/boost/signals2/signal_type.hpp:105:9: error: template argument 1 is invalid ./boost-trunk/boost/signals2/signal_type.hpp:108:31: error: ‘args’ was not declared in this scope ./boost-trunk/boost/signals2/signal_type.hpp:108:67: error: template argument 1 is invalid ./boost-trunk/boost/signals2/signal_type.hpp:111:31: error: ‘args’ was not declared in this scope ./boost-trunk/boost/signals2/signal_type.hpp:111:94: error: template argument 1 is invalid ./boost-trunk/boost/signals2/signal_type.hpp:115:31: error: ‘args’ was not declared in this scope ./boost-trunk/boost/signals2/signal_type.hpp:115:97: error: template argument 1 is invalid ./boost-trunk/boost/signals2/signal_type.hpp:121:13: error: ‘args’ was not declared in this scope ./boost-trunk/boost/signals2/signal_type.hpp:124:11: error: template argument 1 is invalid ./boost-trunk/boost/signals2/signal_type.hpp:128:31: error: ‘args’ was not declared in this scope ./boost-trunk/boost/signals2/signal_type.hpp:128:69: error: template argument 1 is invalid --- [/error] --- I've tried both with last Boost devel version and Boost 1.46.0. Note, I have no problem if one of the two libraries are not used (e.g. if I comment either the first 3 #includes or the last one). Any idea? Thank you very much!! Best, -- Marco

3 3

[config][hp-ux] UTF-8 codecvt_facet.cpp failing on HP-UX
by Beman Dawes 09 Jun '11

09 Jun '11

There is a Filesystem bug report, 5048, that actually affects any library using ...\libs\detail\utf8_codecvt_facet.cpp The problem is line 255: #elif WCHAR_MAX > 0x10000 On HP-UX, WCHAR_MAX is defined like this: # if defined(_INCLUDE_STDCSOURCE_199901) # define WCHAR_MAX UINT_MAX /* max value of an unsigned integer */ # define WCHAR_MIN 0 # else # define WCHAR_MAX (wchar_t)UINT_MAX /* max value of an unsigned integer */ # define WCHAR_MIN (wchar_t)0 # endif An obvious fix is to add #define _INCLUDE_STDCSOURCE_199901 to ...\boost\config\platform\hpux.hpp, and that's been confirmed by the user who originated the bug report. I'm guessing that this fix doesn't work if the user includes <stdint.h> before <boost/config.hpp>, but can't think of anything better. Any objections to committing the change to ...\boost\config\platform\hpux.hpp? --Beman

2 2

[MPL] map size must not be > 19, sunCC vs. gcc
by Bartosz Marcinkiewicz 09 Jun '11

09 Jun '11

Hello, This is on Solaris 11 Express (gcc version 4.3.3, Sun C++ 5.11 SunOS_i386 2010/08/13). When boost::mpl::map<> has more than 19 entries Sun CC does not compile my code. G++ compiles it, even when there are more than 19 entries. [bma@sol11e ~/c++]$ CC -I ~/opt/include -o m3 mpl_map3.cpp && ./m3 "mpl_map3.cpp", line 80: Error: value is not a member of boost::mpl::next<int>. 1 Error(s) detected. [bma@sol11e ~/c++]$ g++ -I ~/opt/include -o m3 mpl_map3.cpp && ./m3 -> 1 -> 2 MPL map size: 20 Both compilers can handle boost::mpl::map with 19 entries. [bma@sol11e ~/c++]$ g++ -I ~/opt/include -o m3 mpl_map3.cpp && ./m3 -> 1 -> 2 MPL map size: 19 [bma@sol11e ~/c++]$ CC -I ~/opt/include -o m3 mpl_map3.cpp && ./m3 -> 1 -> 2 MPL map size: 19 I am attaching code. Is it compiler error in Sun Studio or error in code attached? Thank you for any help. Best regards, bm. -- Bartosz Marcinkiewicz tel. (+48) 516 034 243 e-mail: bma(a)epcon.pl EPCON Sp. z o.o. www.epcon.pl adres: 53-407 Wroclaw, ul. Gajowicka 114/116 tel. (+48) 71 361 75 55 ext. 36 fax (+48) 71 361 74 44

1 1

First Boostcon videos available....
by Marshall Clow 09 Jun '11

09 Jun '11

Keynote: http://blip.tv/boostcon/hans-boehm-threads-and-shared-variables-in-c-0x-key… Chris Kohloff (Monday): http://blip.tv/boostcon/thinking-asynchronously-designing-applications-with… Hartmut (Monday): http://blip.tv/boostcon/phoenix-v3-an-overview-5250984 -- Marshall Marshall Clow Idio Software <mailto:mclow.lists@gmail.com> A.D. 1517: Martin Luther nails his 95 Theses to the church door and is promptly moderated down to (-1, Flamebait). -- Yu Suzuki

10 10

More Boostcon videos available
by Marshall Clow 08 Jun '11

08 Jun '11

Sumant (Mondau): http://blip.tv/boostcon/toward-native-xml-processing-using-multi-paradigm-d… Justin (Tuesday): http://blip.tv/boostcon/intel-s-c-software-transactional-memory-compiler-an… Robert Stewart (Tuesday): http://blip.tv/boostcon/spirit-qi-in-the-real-world-5254335 ( I'm not going to be able to keep up this pace, btw...... ) -- Marshall Marshall Clow Idio Software <mailto:mclow.lists@gmail.com> A.D. 1517: Martin Luther nails his 95 Theses to the church door and is promptly moderated down to (-1, Flamebait). -- Yu Suzuki

2 1