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

[c++now] Perhaps of interest to Boost library users
by Niall Douglas 03 May '15

03 May '15

https://plus.google.com/+nialldouglas/posts/DsAdUV4RgY5 Content of post for convenience: Just finished my C++ Now 2015 presentation with exactly one week to go! Currently 90 slides for a 90 minute long presentation, I had originally been worried I'd be boring but I think, even if I do say so myself, that this will be one of the more interesting talks at C++ Now this year. It reviews *ten* C++ 11/14 only libraries close to entering Boost, three of which in detail (Boost.AFIO, Boost.Fiber, Boost.DI) and one in very great detail (Boost.APIBind), indeed I spend about 45 slides on APIBind as I believe it provides the foundation stone for a fully modular C++ 11 only Boost 2.0 branch with very significant improvements over Boost 1.0. I also draw some sweeping personal general opinions from the ten library review, and I'll try to answer these questions: * Why do these libraries require C++ 11 or 14? From an end user perspective. * Is there a common theme of the most popular C++ 11/14 features used? * Is there a common theme in choice of library design and use of third party libraries? * Are there techniques used in one library which would make a lot of sense to be used in another library, but for some reason are not? * Do these new libraries take notice of one another and integrate themselves well with other libraries, or are they ivory towers? * How many of these forthcoming libraries explicitly seek to contribute to future C++ standardization? Boost only a few years ago was in a state of sickly malaise after the exit of several big names who had dominated it since its inception. Indeed, in 2013 no new libraries had entered in two years, and nobody could say if Boost was going to be the venue for the next generation of C++ 11/14 libraries approaching standardisation or whether some other venue (especially WG21 itself) would take over. Since 2013 *seventeen* new libraries have either entered Boost or entered the formal review queue, almost a third of which are C++ 11/14 mandatory libraries, whilst WG21's efforts to invent new libraries have not panned out well to date. Traffic on the Boost mailing lists which had become anaemic has once again returned to mass-delete-without-reading levels. It looks like Boost is going to remain the place to be for the highest quality C++ libraries, especially those intended for eventual standardisation. The big question is whether the C++ 11/14 mandatory libraries are so different from the 98/03 libraries in design and philosophy that a C++ 11/14 only Boost 2.0 distribution is warranted? Robert Ramey will be presenting on his vision for Boost 2.0 at C++ Now 2015, and he kindly has allowed me to add my ha'pennies worth after he finishes during which I will summarise my empirical findings from the ten library review. So watch out for the recorded videos! -- ned Productions Limited Consulting http://www.nedproductions.biz/ http://ie.linkedin.com/in/nialldouglas/

2 2

[BGL] I think the invariant in ResourceExtensionFunction for r_c_shortest_paths is not correct
by Alberto Maria Santini 02 May '15

02 May '15

The documentation for 1.58.0 (http://www.boost.org/doc/libs/1_58_0/libs/graph/doc/r_c_shortest_paths.html) correctly states that: >>> The implementation is a label-setting algorithm. This means that there must be one or more resources whose cumulated consumption(s) after extension is/are always at least as high as before. This is similar to the Dijkstra algorithm for the SPP without resource constraints where the distance measure must be non-negative. It is sufficient if there is one resource with a non-negative resource consumption along all arcs (for example, non-negative arc lengths or non-negative arc traversal times). >>> But, later on, it says: >>> If ref models the ResourceExtensionFunction concept, and if the type Resource_Container models the ResourceContainer concept, after the call ref( const Graph& g, Resource_Container& new_cont, const Resource_Container& old_cont, graph_traits<Graph>::edge_descriptor ) the expression old_cont <= new_cont evaluates to true. This is because, as stated above, the implementation is a label-setting algorithm. Therefore, the Less-Than operator for ResourceContainer must compare one or more resource(s) whose resource consumption(s) along any arc is/are non-decreasing in order for the algorithm to work properly. >>> Now, saying that "old_cont <= new_cont" is much stricter than it should be. The correct requirement (first quote) is that there is **at least one** resource whose consumption monotonically doesn’t decrease. What "old_cont <= new_cont” requires, is that **all** resources’ consumptions are monotonically non-decreasing. These two things are equivalent iff the label only has one resource. Can someone confirm that my observation makes sense? In this case, I’ll open a ticket and submit a patch for the documentation. Can someone also confirm that this check is *not* enforced in r_c_shortest_paths.hpp? Thanks, AS

1 0

Official Stance on CMake Adoption
by Michael Ainsworth 02 May '15

02 May '15

What is the official stance on CMake adoption? Is if simply a fork in the development, or will all of Boost eventually be built using CMake? Michael

3 3

Re: [Boost-users] [boost] [ANN] "Die Boost C++ Bibliotheken", 2nd edition available (German edition)
by Rüdiger Berlich 01 May '15

01 May '15

Let me just say that the book is excellent! Thanks, Boris, for supplying such a comprehensive source of information on Boost. Best Regards, Ruediger > Am 01.05.2015 um 12:31 schrieb Boris Schäling <boris(a)highscore.de>: > > Just a quick announcement: The 2nd edition of my book "Die Boost C++ Bibliotheken" (German edition) is available. > > - It introduces 72 Boost libraries. > - It contains more than 430 examples. > - The print version has 600 pages. > - There is also a Kindle, Epub and PDF version. > - It's also online: <http://dieboostcppbibliotheken.de/> > > There is hardly any difference between the German and the English language edition. The German language edition is a bit more up-to-date after the English language edition has been published in September last year. But there is no need to read the book again in case you read the English language edition already. > > For those who don't know German: The English language edition remains available online at <http://theboostcpplibraries.com/>. > > Boris > > > _______________________________________________ > Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

1 0

Boost.Cgi Status
by Michael Ainsworth 01 May '15

01 May '15

What is the status of Boost.Cgi by Darren Garvey? It seems that the library hasn't been accepted and it's now 2015 - at least 7 years after last Boost Users mailing list post regarding its status. Michael

2 2

[gsoc15] This year's GSoC students
by Niall Douglas 27 Apr '15

27 Apr '15

Dear Boost, It is our pleasure to announce this year's winning GSoC students and projects: 1. Damian Vicino Topic: Safe Float (boost::safe_numerics<float>) Mentor: Robert Ramey 2. Rajaditya Mukherjee Topic: Boost uBLAS EigenSolver Mentor: David Bellot 3. Anurag Ghosh Topic: Boost Document Library Mentor: Antony Polukhin 4. Nikhar Agrawal Topic: Boost.Fixed-Point Mentor: Paul A. Bristow 5. Benedek Thaler Topic: Enhanced vector and deque containers Mentor: Thorsten Ottosen 6. Louis Dionne Topic: Improved compile-time associative data structures for Hana Mentor: Joel Falcou 7. Jakub Szuppe Topic: Improving Boost.Compute Mentor: Kyle Lutz Congratulations to these students for passing an exceptionally tough contest this year, and our thanks to the above mentors for making themselves available. Finally, thank you Google for another year of funding and to those open source orgs who released duplicate students to us, we appreciate it! Niall and Boris --- Boost C++ Libraries Google Summer of Code 2015 admin https://svn.boost.org/trac/boost/wiki/SoC2015

1 0

Re: [Boost-users] boost examples
by Max 27 Apr '15

27 Apr '15

----- Original Message ----- From: mloskot <mateusz(a)loskot.net> To: boost-users(a)lists.boost.org Subject: Re: [Boost-users] boost examples Date: 2015-04-24 18:01 mloskot wrote > On 23 April 2015 at 11:44, Trek < > trek830@ > > wrote: >> Is there a central link where all official boost examples are hosted? > > No. You may check http://theboostcpplibraries.com/ for unofficial ones. ----- -- Mateusz Loskot, http://mateusz.loskot.net Charter Member of OSGeo, http://osgeo.org Member of ACCU, http://accu.org The book, Beyond the C++ Standard Library, also used to be a very good starting point of my understanding the boost libraries. Max

1 0

[Pyhton] (re)binding class instance in python object
by Tamas.Ruszkai＠leica-geosystems.com 27 Apr '15

27 Apr '15

Hi, I have a plain C++ class which I want to exposed to python. The pyhton object accesses the C++ class via smart pointer. class Sensor : public boost::enable_shared_from_this<Sensor> { public: Sensor(int port); ~Sensor() void Open(); void SendCommand(const std::string& command); void Close(); private: int _port; }; auto sensorClassTypeObject = boost::python::class_<Sensor, boost:: shared_ptr<Sensor>>("Sensor", boost::python::init<int>()) .def("Open", &Sensor::Open) .def("SendCommand", &Sensor::SendCommand) .def("Close", &Sensor::Close); The question is: if I already have one instance of the C++ class (in a dependency injection container) then how can I create in the C++ part an instance of the python wrapper so that it does not create a new instance of Sensor, but uses my existing Sensor instance? Then I would put this instance into one of the python modules (this part would be made in C++) inside the python interpreter to be able to access it via script. Something like that: shared_ptr<Sensor> sensorInstance = DIContainer.Resolve<Sensor>(); boost::object pythonInstance = sensorClassTypeObject(sensorInstance); pythonInterpreterWrapper.AddToModule("__main__", pythonInstance); Best Regards, Tamas

1 0

[msm] Event does not get processed after exiting an interrupt state
by Chrisopher Vogl 27 Apr '15

27 Apr '15

Hi all! I got a state machine with 2 regions. The second region has two states ("Run" and "Interrupt"), one is an interrupt state. The region starts in the non interrupt state, just like in the MSM example "Orthogonal regions, terminate state, event deferring". The event called Ev_Resume leads to a transition from state Interrupt back to state Run. In on_entry() of Run I fire an event which never gets processed. Is at that time the FSM still interrupted, i.e. event processing still blocked? Christopher

2 1

Re: [Boost-users] [boost] [test] boost.test owner unresponsive to persistent problems for multiple years
by Paul A. Bristow 24 Apr '15

24 Apr '15

> -----Original Message----- > From: Boost [mailto:boost-bounces@lists.boost.org] On Behalf Of Richard > Sent: 17 March 2015 17:20 > To: boost(a)lists.boost.org > Subject: Re: [boost] [test] boost.test owner unresponsive to persistent problems for multiple years > > Another quarter has gone by and another Boost release is prepared for beta and I see *no* changes to > Boost.Test listed in the release notes.> > So here we are, waiting another 3 months for *simple* changes to be reflected in this library while the > maintainer insists that the fixes are available, yet they continue not to be deployed into any shipping release. It is indeed really deplorable that we don't have Boost.Test updates in a release. Though in mitigation I note that we are really struggling to get out a GIT-driven release at all, (and some of the new troubles are with asynch-exceptions in Boost.Test. See this thread Re: [boost] Running b2 on develop needs asynch-exceptions=on.) However, Richard's hard work on improved docs have been refined and completed by Raffi Enficiaud. Many bug fixes are in the develop branch (including some very new and prompt bug fixes, for example https://svn.boost.org/trac/boost/ticket/11054) I also understand that the fear of Boost.Test updates causing trouble with other libraries, and worse with users test, is a cause for caution in a release version. So it would be really good if *lots more users could download and use the version in the Boost.Test develop branch* so we can get current user feedback. (My experience is good using this, and the test matrix is also good). You can download just Boost.Test either using GIT or as a zip from https://github.com/boostorg/test/tree/develop and substitute it for the contents of the current release modular-boost/libs/test folder (Don't forget to rebuild the Boost.Test library unless you only use the included version). If nothing else, you will find Raffi's shiny new Boostbook docs finally allow you to find what you want to know without frustration. I'd very much like to see this new docs version in the 1.58 documentation release (despite any new features not being in the code release). Paul --- Paul A. Bristow Prizet Farmhouse Kendal UK LA8 8AB +44 (0) 1539 561830

5 5

[log] Possible to explicitly disable AVX2?
by Jason Roehm 24 Apr '15

24 Apr '15

I am building Boost.Log on a relatively new machine whose compiler and CPU both support the AVX2 instruction set. Boost.Log detects this at build time and configures its compiler flags accordingly to use the AVX2 instruction set. I'd like to use the resulting binaries, however, on older systems that do not have AVX2 support. Is it possible to explicitly disable this at build time? I didn't see any mention of it in the documentation, and my attempts to parse the Jamfile's intentions were not successful. Thank you. Jason

1 0

[nowide] Cannot end input using ^Z/^D on Windows
by Yuan Yao 24 Apr '15

24 Apr '15

3 2

boost examples
by Trek 24 Apr '15

24 Apr '15

Is there a central link where all official boost examples are hosted? thanks,

3 2

Boost.Thread Continuation future with executor blocking in destructor
by Konrad Zemek 23 Apr '15

23 Apr '15

Hi, When creating a continuation future with future<>::then(), and the policy is boost::launch::async, the created future's destructor blocks, like a future created by boost::async, and probably for the same reasons (as outlined by the standardization committee in n3679). The same behavior is coded into continuation futures' destructor when using an executor. Why is it needed here? The situation with executor-run continuations is a little different, as even when the future is destroyed, the lifetime of running job is still bound to the executor. My use case: I've coded a network communication stack that returns to the caller a future<NetworkResponse>. I've previously used std::async(std::launch::deferred, ...) to transform the future's content before returning it to the caller (bytes -> protobuf -> internal object), and I consider such manipulation of a future value to be a very powerful feature. I've used std::launch::deferred to reduce the number of running threads, but the downside is that the client can't wait for the future value with a timeout. On the other side of the spectrum is std::launch::async, which would run a new thread - per pending communication - that would do little more than block. boost::future<>::then is a fantastic fit for my use case, as I can use my boost::asio::io_service as an executor and let my communication stack's threads do the work without having them block on future.get() to first retrieve the result. The caller can then call future<>::wait_for() to wait for the reply with a custom timeout. This being network communication, though, a reply message may never arrive - the corresponding promise will eventually be destroyed by the stack, but I can't have users block on the future's destructor until that happens, after they have already decided that the reply is no longer worth waiting for. Please advise if there's a workaround for this behavior that doesn't involve me distributing a custom version of Boost.Thread with my binaries? :) Konrad

5 6

[asio] What are the enhancements for Asio in 1.58
by Aaron Levy 22 Apr '15

22 Apr '15

I am sorry if this has been already asked. Wanted to get a head start on Asio 1.58, having heard that it's a significantly enhanced library. -- Aaron Levy aaron.levy(a)yandex.com

3 2

[multi_index] Announcing preview of ranked indices
by Joaquin M Lopez Munoz 22 Apr '15

22 Apr '15

Hi, It's been 11 years since first proposed, but here it is at last. I'm releasing a prereview of ranked indices for Boost.MultiIndex: struct person { int age; std::string name; }; typedef multi_index_container< person, indexed_by< ranked_non_unique<member<person,int,&person::age> >, ordered_non_unique<member<person,std::string,&person::name>>>> multi_t; multi_t m={{40,"Joe"},{25,"Jill"},{30,"Kurt"},{32,"Sue"}}; auto it=m.emplace(31,"Maggie").first; std::cout<<m.rank(it)<<" persons younger than "<<it->name<<"\n"; auto n=m.size()/2; std::cout<<n<<" persons are less than "<<m.nth(n)->age<<" years old\n"; Ranked indices add a bunch of extra rank-related capabilities to the interface of ordered indices. The rank of an element is its numerical position in the index. nth(i) returns an iterator to the element with rank i, whereas rank(it) is the inverse operation. Both execute in log(n) time. Additionally, the member functions find_rank lower_bound_rank upper_bound_rank equal_range_rank range_rank behave as their "_rank"-less counterparts except they return ranks rather than iterators. One drawback of ranked indices wrt ordered indices (other than their slower perfomance and higher memory consumption) is that deletion of elements is log(n) (in ordered indices it is constant time). Download lib preview: http://tinyurl.com/kemwk8q Tutorial: http://tinyurl.com/q36c9f7 Reference: http://tinyurl.com/pvoulgr For people interested in this, I'd appreciate if you could download the source, copy it on top of Boost 1.58 (1.57 should work as well), play a bit with the new feature and report your results. Also, opinions on naming are welcome. Target version for releasing is Boost 1.59, so we have plenty of time to discuss the design before it is final. Thank you, Joaquín M López Muñoz Telefónica

1 0

Which is the effective way for data transfer for Interprocess shared memory ?
by Kalyan Chakravarthy Gudimella 22 Apr '15

22 Apr '15

Hi, I am currently evaluating boost interprocess library for using shared memory. I found two ways to transfer complex data structures. 1. Creating data structures directly in shared memory. E.g. http://www.boost.org/doc/libs/1_58_0/doc/html/interprocess/allocators_conta… class complex_data{ int id_; char_string char_string_; int_vector_vector int_vector_vector_; public: //Since void_allocator is convertible to any other allocator<T>, we can simplify //the initialization taking just one allocator for all inner containers. complex_data(int id, const char *name, const void_allocator &void_alloc) : id_(id), char_string_(name, void_alloc), int_vector_vector_(void_alloc) {} //Other members...}; 2. Serializing data into bufferstream and deserializing it. E.g.http://www.boost.org/doc/libs/1_58_0/doc/html/interprocess/streams.html My initial trial shows that streaming to bufferstream took 7 times more time than using direct shared structures. Can you please help me to choose the effective way between these two approaches? thanks, Kalyan

1 0

[MPL][Config] Can no longer pre-generate headers for MPL-containers with more than 50 elements.
by Deniz Bahadir 22 Apr '15

22 Apr '15

Hello everyone, While playing around with the current Boost release-candidate (1.58.0-RC3) I noticed that pre-generating MPL-headers for MPL-containers with more than 50 Elements seems no longer possible. Up till at least Boost 1.55.0 I was able to pre-generate these MPL-headers using the python-scripts from Boost-source ([1]) with the help of a bash-script which I already described here [2]. Internally, these python-scripts seem to use GCC to generate these MPL-headers. They also seem to partially circumvent Boost.Config by using custom user.hpp files located in subdirectories of [3] and defining BOOST_NO_CONFIG and some other macros therein. However, this does not seem to please Boost.Config so that compiler-errors ([4]) occur. Defining the macro BOOST_CONFIG_ALLOW_DEPRECATED as [4] suggests seems not to help. It just silences the warnings but does not generate proper files. (It actually erases the content of almost all header-files in subdirectories of [5].) Does anyone know how to workaround or better fix this? Or does anyone know who to talk to, to get these problems fixed? Also, I would suggest, that documentation for these python-scripts are added to the documentation of Boost.MPL. Best regards, Deniz References: [1] ${BOOST_SOURCE}/libs/mpl/preprocessed/ [2] http://stackoverflow.com/a/20660264/3115457 [3] ${BOOST_SOURCE}/libs/mpl/preprocessed/include/ [4] #error "You are using a compiler which lacks features which are now a minimum requirement in order to use Boost, define BOOST_CONFIG_ALLOW_DEPRECATED if you want to continue at your own risk!!!" [5] ${BOOST_SOURCE}/boost/mpl/aux_/preprocessed/

3 5

[SPAM] from: giridhar
by giridhar 22 Apr '15

22 Apr '15

Greetings boost http://findislamicservices.com/reduce.php?silver=aswydt7xv2r0enm3 giridhar Sent from my iPhone

1 0

sliding window containing n seconds of data
by Lane 22 Apr '15

22 Apr '15

I'm looking for a way to examine an array of object structures for an adjustable window of n seconds. My application reads objects at a high rate (which I'll have to throttle), and I'd then like to put them onto an array. My thinking is to assign a timer on each object that is initialized to n seconds and once the timer expires, the object is removed from the array, but while it's on the array, I'll iterate over the objects looking for the info I need. I put this out on the boost list as I suspect boost has the mechanisms to handle this and someone could suggest a good way of going about it. Any help much appreciated.

5 5

1.58.0 rc3 is now available
by Marshall Clow 22 Apr '15

22 Apr '15

Release candidate files for 1.58.0 are available at: http://boost.cowic.de/rc/ * Since rc1, the changes that were causing the serialization libraries to fail to build on MSVC have been reverted. * Since rc2, the documentation for bind, and mem_fun. No code changes. As always, the release managers would appreciate it if you download the candidate of your choice and give building it a try. Please report both success and failure, and anything else that is noteworthy. This helps ensure the candidates build OK before we push them out to SourceForge. The files (and associated md5s) are: MD5 (boost_1_58_0_rc3.7z) = f7255aeb692c1c38fe761c32fb0d3ecd MD5 (boost_1_58_0_rc3.tar.bz2) = b8839650e61e9c1c0a89f371dd475546 MD5 (boost_1_58_0_rc3.tar.gz) = 5a5d5614d9a07672e1ab2a250b5defc5 MD5 (boost_1_58_0_rc3.zip) = b0605a9323f1e960f7434dbbd95a7a5c Thanks! -- The release managers P.S. Release notes are here: http://www.boost.org/users/history/version_1_58_0.html

5 7

Re: [Boost-users] [IPC Named Mutex]
by Lorenzo Trivelli 21 Apr '15

21 Apr '15

Hi Tom, I tried your solution and it worked, and I'm stil wondering about the rationale behind the bootup time usage ;-) I also submitted a bug ticket, #11194, but track took a very long time to create it and add the test case, and in the end I found that: 1) a bug with ticket !11195 was also created 2) the test case was attached four times :-( can anyone please clean it up as is is a quite ugly ? ;-) regards Lorenzo

1 0

Re: [Boost-users] [Serialization / 1.58] wrong number of entries in std::vector<boost::tuple<double, double, double> >
by Rüdiger Berlich 21 Apr '15

21 Apr '15

Hi Robert, I have submitted trac item 11214 and attached the test case. The trac web interface somehow froze for me upon submission (Safari / MacOS), and I apparently have pressed the submit button more than once, trying to submit the track item. So it appears as if more than one ticket has been created (11215 - 11218), which I do not seem to be able to delete. Please ignore these -- apologies. Best Regards, Beet > > ouch! - make a trac item for this! > > Robert Ramey > > > > -- > View this message in context: http://boost.2283326.n4.nabble.com/Serialization-1-58-wrong-number-of-entri… > Sent from the Boost - Users mailing list archive at Nabble.com. > _______________________________________________ > Boost-users mailing list > Boost-users(a)lists.boost.org > http://lists.boost.org/mailman/listinfo.cgi/boost-users >

1 0

Web doc 1.58 Categorized error
by Larry 20 Apr '15

20 Apr '15

I tried to look at the 1.58 doc on the website and clicking on Categorized gave me under the headings By Category Unknown category: Data structures Apparently missing something. “All”, “Standard Proposals” etc seem to work. Larry

2 1

[Serialization / 1.58] wrong number of entries in std::vector<boost::tuple<double, double, double> >
by Rüdiger Berlich 20 Apr '15

20 Apr '15

Hi, I’ve just installed Boost 1.58 and ran into the following problem: I have a class with a std::vector<boost::tuple<double, double, double> >, that is supposed to always have at least one entry, but may have more. Hence, in the default constructor, I initialize the std::vector with one entry. The class is also serializable. Up until Boost 1.57, a de-serialized object always had the correct number of entries in the vector. As of 1.58, I am getting one entry too much, which probably means that Boost.Serialization simply attaches entries to the vector of a default-constructed class, but doesn’t clear it first. This happens in C++11-Mode, both on Ubuntu 14.10 (gcc 4.9.1) as well as on MacOS X Yosemite . It DOES NOT happen with a std::vector<double> . Below is a full test case that replicates the problem in a simplified environment. For Boost 1.57 the program returns: t_vec.size() = 1 t_vec2.size() = 1 For Boost 1.58 the answer is t_vec.size() = 1 t_vec2.size() = 2 Best Regards, Beet /******************************************************/ #include <boost/archive/xml_oarchive.hpp> #include <boost/archive/xml_iarchive.hpp> #include <boost/serialization/nvp.hpp> #include <boost/serialization/vector.hpp> #include <boost/tuple/tuple.hpp> #include <iostream> #include <vector> #include <sstream> namespace boost { namespace serialization { template<typename archive, typename T0, typename T1, typename T2> void serialize(archive & ar, boost::tuple<T0, T1, T2> & tpl, unsigned int) { using namespace boost; using boost::serialization::make_nvp; ar & make_nvp("tpl_0", boost::get<0>(tpl)) & make_nvp("tpl_1", boost::get<1>(tpl)) & make_nvp("tpl_2", boost::get<2>(tpl)); } } /* namespace serialization */ } /* namespace boost */ int main() { std::vector<boost::tuple<double, double, double> > t_vec, t_vec2(1); std::ostringstream oss; t_vec.push_back(boost::tuple<double, double, double>(1.,2.,3.)); std::cout << "t_vec.size() = " << t_vec.size() << std::endl; // save data to archive { boost::archive::xml_oarchive oa(oss); // write class instance to archive oa & BOOST_SERIALIZATION_NVP(t_vec); // archive and stream closed when destructors are called // Data is now in oss } // load data from archive { std::istringstream iss(oss.str()); boost::archive::xml_iarchive ia(iss); // read class state from archive ia & BOOST_SERIALIZATION_NVP(t_vec2); // archive and stream closed when destructors are called } std::cout << "t_vec2.size() = " << t_vec2.size() << std::endl; }

2 1