December 2015 - Boost-users

Boost 1.60.0 has been released
by Marshall Clow 23 Dec '15

23 Dec '15

Release 1.60.0 of the Boost C++ Libraries is now available. These open-source libraries work well with the C++ Standard Library, and are usable across a broad spectrum of applications. The Boost license encourages both commercial and non-commercial use. This release contains one new library and numerous enhancements and bug fixes for existing libraries. New Libraries * VMD: Variadic Macro Data library, from Edward Diener For details, including download links, see http://www.boost.org/users/news/version_1.60.0 You can also download directly from SourceForge: http://sourceforge.net/projects/boost/files/boost/1.60.0/ Tom Kent has made pre-built windows binaries, and they are available at: https://sourceforge.net/projects/boost/files/boost-binaries/1.60.0/ To install this release on your system, see http://www.boost.org/doc/libs/release/more/getting_started/index.html Thanks, --The Boost release team Vladimir Prus Rene Rivera Marshall Clow Eric Niebler Daniel James Beman Dawes

4 4

Race in boost::condition_variable::wait
by Chris Stankevitz 23 Dec '15

23 Dec '15

Hi, Regarding boost::condition_variable::wait (listed below with an obvious modification), It seems to me that there is a race condition because the following operations are not atomic: 1) unlocking of "m" 2) waiting on "cond". Can someone explain to me why boost::condition_variable::wait is not racing the user code in the example below? Thank you, Chris === inline void condition_variable::wait(unique_lock<mutex>& m) { int res=0; { thread_cv_detail::lock_on_exit<unique_lock<mutex> > guard; detail::interruption_checker check_for_interruption(&internal_mutex,&cond); guard.activate(m); #ifdef ENCOURAGE_DEADLOCK boost::this_thread::sleep_for(boost::chrono::seconds(100)); #endif do { res = pthread_cond_wait(&cond,&internal_mutex); } while (res == EINTR); } } === #define ENCOURAGE_DEADLOCK #include <boost/thread.hpp> int main() { boost::mutex Mutex; boost::condition_variable Condition; bool Quit = false; boost::thread Thread([&] { boost::unique_lock<boost::mutex> Lock(Mutex); while (!Quit) { Condition.wait(Lock); } }); boost::this_thread::sleep_for(boost::chrono::seconds(1)); boost::lock_guard<boost::mutex> Lock(Mutex); Quit = true; Condition.notify_all(); return 0; }

1 0

Not for publishing
by Soul Studios 22 Dec '15

22 Dec '15

To whoever it is that's been blocking my posts to the mailing list - with the exception of this email, please stop. I have chronic fatigue syndrome, and I feel like I'm fighting an uphill battle against the lack of documentation around actually doing documentation for boost. I would've understood if I'd said anything reasonably offensive, but I don't think I did.

1 0

Software Freedom Conservancy Support
by Jon Kalb 22 Dec '15

22 Dec '15

This is not a technical message, but instead focuses on Boost Administration. Software Freedom Conservancy in a Financial Fight for its Life Software Freedom Conservancy, the non-profit home of Boost is facing an existential threat. Historically, Conservancy has been funded by corporate donations. This works great when it works, but a management shake-up, change in priorities, or a bad quarter at a large donor can derail the organization¹s programs. It has become obvious that Conservancy must move to a Supporter-driven funding model, where individuals who believe in Conservancy¹s mission of supporting open source projects, such as Boost, make a personal commitment to support that effort. Conservancy's fundraiser, found at https://sfconservancy.org/supporter/, is centered around asking individuals who care about the work that Conservancy does to donate $120/year, or just $10/month, to support the organization. Conservancy seeks about 2,500 Supporters to continue its current work, and 750 Supporters to continue a ³bare minimum² of services to its projects. What does this mean to Boost? Boost receives a myriad of important services (see https://sfconservancy.org/members/services/ ) from Conservancy. We¹ve relied on Conservancy since 2007 to provide them. Here are some of the things that Conservancy does for Boost that are in the ³bare minimum² category: * Dealing with all the financial and contractual details regarding our annual C++ Now conference. It was BoostCon that brought Boost and Conservancy together. C++Now 2016 will the the tenth annual BoostCon and it is a tribute to what we¹ve accomplished working together. * Handle our Google Summer of Code payments and travel, including handling all financial transactions with Google to receive and then disperse the money. * Coordinating travel reimbursements for Boost contributors (both developers and students) to attend conferences. * Receiving and tracking all donations to Boost. * Keeping the non-profit status for us so that we don't have to form our own org and file our own paperwork. As the Chair of the Boost Steering Committee, I cannot overstate how important this is to our existence as an organization. Here are some of the services Boost has received in the wider category: * Dealing with questions from the community (including license-governing bodies like OSI) about Boost's license. * Discussing the issue of copyright governance and contributions with large corporations who want their employees to contribute to Boost but need legal assurances or have questions relating to Boost's license or contribution policies. * Reviewing and handling licensing terms of third-party software that relate to Boost in some way, and giving us feedback and answers about issues related to it. * General legal consulting on issues that we run into as an open source non-profit. As you can see, Boost really relies on the important work of Conservancy. Support from Boost Boost, like all of Conservancy¹s projects, contributes 10% of our revenue to Conservancy¹s general fund. (The other 90% is maintained and administered by Conservancy at the direction of the Boost Steering Committee to advance the Boost project.) But that 10% of revenue from all Conservancy¹s member projects doesn¹t add up to enough to even employee one full time person, let alone the already overworked staff of three that Conservancy has. Donors must make up the difference. When Conservancy launched its fundraising effort, I sent out a request to past BoostCon/C++Now attendees asking for support and I¹m proud and delighted to say that a number of Boost community members have already stepped up. One told me that her company will be matching her support. Boost is one of Conservancy¹s oldest and largest projects, so it is right that Boost community members make up a fair proportion of its supporters. I encourage you to choose Conservancy as one of your charitable gifts this holiday season. (Conservancy is a 501(c)(3) charity incorporated in New York, and donations are typically deductible on U.S. taxes.) Please visit: https://sfconservancy.org/supporter/ If you are able to join the others from the Boost community that are making this gift, please let me know. I¹d like to personally thank you. Jon Kalb Chair, Boost Steering Committee

1 0

boost.test: Run multiple tests
by Udo Steinbach 22 Dec '15

22 Dec '15

Hello, boost:test is a great tool. Now I have boost updated to v1.60 and can't run more than one tests at once. System is MinGW on Windows. Given suite_1 test_1_1 suite_1_2 test_1_2_1 and many more, all enabled by default, with v1.59 I could run these both test and exclude all others with --run_test=suite_1/test_1_1,suite_1/suite_1_2/test_1_2_1 respectively BOOST_TESTS_TO_RUN, but with v1.60 the result is "Test setup error: no test cases matching filter or all test cases were disabled". With --run_test=suite_1/test_1_1,suite1_2/test_1_2_1 only test_1_2_1 is executed. Same result with BOOST_TEST_RUN_FILTERS=suite_1/test_1_1,suite1_2/test_1_2_1 . With multiple --run_test= all is fine. But how do I execute multiple tests via variable? bye Udo -- Fahrradverkehr in Deutschland: http://radwege.udoline.de/ GPG: A245 F153 0636 6E34 E2F3 E1EB 817A B14D 3E7E 482E

2 2

Johnson all pair shortest path and output in a C array
by Paolo Bolzoni 22 Dec '15

22 Dec '15

Dear list, I need to get the all pair shortest path in a C array, to pass it to a java program. However johnson all pair shortest path[1] requires a BasicMatrix[2] as output, so a simple array won't do. At the moment I used few classes (see Matrix and MatrixR[3]) as wrapper around a std::vector and it works fine, but I was wondering if I am missing a simpler solution. Is there? Yours faithfully, Paolo [1]http://www.boost.org/doc/libs/release/libs/graph/doc/johnson_all_pairs_sh… [2]http://www.boost.org/doc/libs/release/libs/graph/doc/BasicMatrix.html [3] struct Matrix { Matrix(std::vector<int64_t>& storage, size_t rows, size_t columns) : m_(storage), columns_(columns) { m_.clear(); m_.resize(rows * columns_, int64_t{0}); } int64_t& get(size_t row, size_t column) { return m_[row * columns_ + column]; } std::vector<int64_t>& m_; size_t columns_; }; struct MatrixR; struct MatrixRR { MatrixRR(MatrixR const& m, size_t column) : m_(m), column_(column) {} int64_t& operator[] (size_t row) const; MatrixR const& m_; size_t column_; }; struct MatrixR { MatrixR(Matrix& m) : m_(m) {} MatrixRR operator[] (size_t column) const { return MatrixRR(*this, column); } Matrix& m_; }; int64_t & MatrixRR :: operator[] (size_t row) const { return m_.m_.get(row, column_); }

1 0

Re: [Boost-users] [container] Boost 1.60 Release link error
by Hickman, Steve (AdvTech) 21 Dec '15

21 Dec '15

I'm using VS2012 on Win 7 64bit building 32bit executables. --- Steve H. From: Hickman, Steve (AdvTech) Sent: Monday, December 21, 2015 11:14 AM To: 'boost-users(a)lists.boost.org' <boost-users(a)lists.boost.org> Subject: [container] Boost 1.60 Release link error Getting a linker error with the Windows version of Boost on release only (not debug) using v 1.60. This did not show up with 1.58 (I skipped 1.59 so I have no comment on it). I'm using the prebuilt libraries: 8> Creating library D:\GITRepos\fd_x-current_feature\source\Devices\MockDevice\Test\Release\MockTest.lib and object D:\GITRepos\fd_x-current_feature\source\Devices\MockDevice\Test\Release\MockTest.exp 8>MockDevice.lib(Mock.obj) : error LNK2001: unresolved external symbol "public: void __thiscall boost::container::container_detail::has_member_function_named_construct<class boost::container::new_allocator<char> >::BaseMixin::construct(void)" (?construct@BaseMixin@?$has_member_function_named_construct@V?$new_allocator@D@container@boost@@@container_detail@container@boost@@QAEXXZ) 8>MockDevice.lib(Mock.obj) : error LNK2001: unresolved external symbol "public: void __thiscall boost::container::container_detail::has_member_function_named_construct<class boost::container::new_allocator<struct boost::container::container_detail::tree_node<struct std::pair<class boost::container::basic_string<char,struct std::char_traits<char>,class boost::container::new_allocator<char> > const ,struct boost::interprocess::ipcdetail::ref_count_ptr>,void *,0,1> > >::BaseMixin::construct(void)" (?construct@BaseMixin@?$has_member_function_named_construct@V?$new_allocator@U?$tree_node@U?$pair@$$CBV?$basic_string@DU?$char_traits@D@std@@V?$new_allocator@D@container@boost@@@container@boost@@Uref_count_ptr@ipcdetail@interprocess@3@@std@@PAX$0A@$00@container_detail@container@boost@@@container@boost@@@container_detail@container@boost@@QAEXXZ) 8>D:\GITRepos\fd_x-current_feature\source\Devices\MockDevice\Test\Release\MockTest.exe : fatal error LNK1120: 2 unresolved externals Steve Hickman System Architect, Human Machine Collaboration Honeywell | Aerospace Office: 480-236-8367 steve.hickman(a)honeywell.com<mailto:steve.hickman@honeywell.com> https://in.honeywell.com/sites/aero/Pages/Home.aspx

1 0

segfault using boost::thread_specific_ptr
by Michael P. Soulier 21 Dec '15

21 Dec '15

Hello, I’m trying to implement a thread-safe logger that uses a boost::thread_specific_ptr to a std::stringstream as a logging buffer. static boost::thread_specific_ptr<std::stringstream> tls_buffer; class MLoggerHandler { public: MLoggerHandler(boost::mutex& mutex, std::ostream& ostream); ~MLoggerHandler(); void setLevel(int level); template <class T> // For handling << from any object. MLoggerHandler& operator <<(T input) { std::cerr << "Adding to buffer" << std::endl; if (m_start) { *tls_buffer << localDateTime() << " " << input; <================= SEGFAULT m_start = false; } else { *tls_buffer << input; } std::cerr << "On buffer: " << tls_buffer->str() << std::endl; return *this; } // For handling std::endl std::ostream& operator <<(std::ostream& (*f)(std::ostream&)) { std::cerr << "Flushing buffer" << std::endl; m_start = true; std::ostream& rv = f(m_ostream); m_mutex.lock(); m_ostream << tls_buffer->str(); m_mutex.unlock(); tls_buffer->str(""); return rv; } private: // A flag to indicate the beginning of a new line. bool m_start = true; // A mutex passed in from the main logger for synchronization. boost::mutex& m_mutex; // The logging level. int m_level; // The output stream. std::ostream& m_ostream; // Return the current date and time as a localized string. const std::string localDateTime(); }; The initial idea, was to flush the buffer and write when a terminating std::endl was encountered. But, just trying to dereference the pointer and write to the std::stringstream is crashing. Do I need to initialize the std::stringstream, or is it initialized in constructor? Thanks, Mike

2 3

Problems building with Windows 10.0.10586.0 SDK
by Isaac Lascasas 21 Dec '15

21 Dec '15

I have just downloaded boost 1.60 for a regular update, I have found some kind of build problem with my environment (vs2015 update 1 with Windows SDK 10.0.10586.0). I usually just do this to get a build: call bootstrap b2 headers bjam --build-type=complete --toolset=msvc-14.0 -j13 address-model=64 stage I have never required vcvarsall.bat as boost build allways finds its way. However today it wasn't able to find my windows headers so I manually did this: %comspec% /k ""C:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\vcvarsall.bat"" x86_amd64 10.0.10586.0 Even so I found that boost was looking for sdk 10.0.10240.0 (an older one). In my system I still had that sdk directory, however it only contained some UCRT stuff. I uninstalled deliberately that version with my visual studio setup some days ago to upgrade to the new SDK. Until now I didn't found any problems. Maybe you should check this out a bit, I can't figure out if this is a problem of my environment, microsoft or boost build. Bringing 10240 back allows me to build boost 1.60 without any problems. My biggest concern is that boost is building against 10240 when 10586 is available and both are installed, even after an explicit vcvars call to stick to 10586. Best, Isaac Lascasas.

1 0

Duplicate function in composite_key.hpp ?
by John M. Dlugosz 21 Dec '15

21 Dec '15

I'm getting E:\C++Libs\boost_1_59_0\boost/multi_index/composite_key.hpp(638): error C2535: 'boost::multi_index::composite_key_result<boost::multi_index::composite_key<const vst::Foo1::Bar2::Baz3,boost::multi_index::const_mem_fun<vst::Foo1::Bar2::Baz3,UID,UID vst::Foo1::Bar2::Baz3::owner(void) const>,boost::multi_index::member<vst::Foo1::Bar2::Baz3,const uint32_t,pointer-to-member(0x18)>,boost::tuples::null_type,boost::tuples::null_type,boost::tuples::null_type,boost::tuples::null_type,boost::tuples::null_type,boost::tuples::null_type,boost::tuples::null_type,boost::tuples::null_type>> boost::multi_index::composite_key<const vst::Foo1::Bar2::Baz3,boost::multi_index::const_mem_fun<vst::Foo1::Bar2::Baz3,UID,UID vst::Foo1::Bar2::Baz3::owner(void) const>,boost::multi_index::member<vst::Foo1::Bar2::Baz3,const uint32_t,pointer-to-member(0x18)>,boost::tuples::null_type,boost::tuples::null_type,boost::tuples::null_type,boost::tuples::null_type,boost::tuples::null_type,boost::tuples::null_type,boost::tuples::null_type,boost::tuples::null_type>::operator ()(const boost::reference_wrapper<const Class> &) const': member function already defined or declared with [ Class=vst::Foo1::Bar2::Baz3 ] (compiling source file Bar2\Delta.cpp) (actual names from my code have been sanitized) Using Boost 1.59 and VS2015. The function it points to is result_type operator()(const reference_wrapper<value_type>& x)const but the error message looks like it is duplicating the previous function in the file, result_type operator()(const reference_wrapper<const value_type>& x)const and the chain of instantiation comes through composite_key.hpp, detail/ord_index_args.hpp(46), mpl/eval_if.hpp(41), detail/ord_index_args.hpp(70), ordered_index.hpp(65) with [ Arg1=boost::multi_index::tag<UID,boost::mpl::na,boost::mpl::na,boost::mpl::na,boost::mpl::na,boost::mpl::na,boost::mpl::na,boost::mpl::na,boost::mpl::na,boost::mpl::na,boost::mpl::na,boost::mpl::na,boost::mpl::na,boost::mpl::na,boost::mpl::na,boost::mpl::na,boost::mpl::na,boost::mpl::na,boost::mpl::na,boost::mpl::na>, Arg2=boost::multi_index::composite_key<const vst::Foo1::Bar2::Baz3,boost::multi_index::const_mem_fun<vst::Foo1::Bar2::Baz3,UID,UID vst::Foo1::Bar2::Baz3::owner(void) const>,boost::multi_index::member<vst::Foo1::Bar2::Baz3,const uint32_t,pointer-to-member(0x18)>,boost::tuples::null_type,boost::tuples::null_type,boost::tuples::null_type,boost::tuples::null_type,boost::tuples::null_type,boost::tuples::null_type,boost::tuples::null_type,boost::tuples::null_type>, Arg3=boost::mpl::na ] What's going on here? This worked previously with Boost 1.55 and VS2010.

1 0

[serialization] How to serialize an atomic<> value?
by Merrill Cornish 20 Dec '15

20 Dec '15

I am running Boost_1_60_0 on Windows 10 64bit on MinGW64: I have this class data member I want to serialize: std::atomic< std::time_t > mUpdate; Since I'm on a 64-bit machine, serialization recognizes that as atomic< long long int >. However, given ar & mUpdate; serialization says it does not know how to serialize atomic< long long int >. My solution, which compiles cleanly, is std::time_t updateTime = mUpdateTime.load(); ar & updateTime; mUpdateTime = updateTime; // assign is atomic My reasoning is that on load (ar => mUpdateTime), the first line is useless, but harmless. Similarly, on save (ar <= mUpdateTime), the last line is redundant, but also harmless. Of course, the application is not executing during serialization and de-serialization. As I said, this compiles without error or warning; but I worry that I have missed something. Any thoughts? Merrill Cornish Once I see how to get the above working, my next project will be std::atomic< std::shared_ptr< boost::variant<...> > mValue;

2 2

Why does BOOST_TEST() treat std::string as a collection?
by Brad Spencer 20 Dec '15

20 Dec '15

Why does BOOST_TEST() compare std::string values as collections instead of scalars (by default)? It's odd that the following produce different output, for example: BOOST_TEST(std::string("a") == "b"); BOOST_TEST(std::string("a") == std::string("b")); This will output: test.cc(6): error: in "test": check std::string("a") == "b" has failed [a != b] test.cc(7): error: in "test": check std::string("a") == std::string("b") has failed Note the lack of "[a != b]" on the second case. Also note that BOOST_CHECK_EQUAL() would emit as in the first form. This seems to be because std::string is_forward_iterable, and thus falls into the collection_comparison_op.hpp comparators instead of the scalar comparators. Is this intentional or an oversight? As a simple fix, adding a condition to the partial specialization of the collection_comparison_op.hpp comparators that specifically excludes std::string is enough to have std::string treated as scalars. For example, to handle std::string but not std::wstring with C++11 support: template<typename Lhs,typename Rhs> struct name<Lhs,Rhs,typename boost::enable_if_c< unit_test::is_forward_iterable<Lhs>::value && unit_test::is_forward_iterable<Rhs>::value && !(std::is_same<typename std::decay<Lhs>::type, std::string>::value || std::is_same<typename std::decay<Rhs>::type, std::string>::value) >::type (The above is an example and not a complete change supporting std::wstring and C++98.) -- Brad Spencer

2 1

Boost.Serialization broken for wchar_t
by gast128 19 Dec '15

19 Dec '15

following code used to work in boost 1.57 void f() { std::wstring wstr = L"\u20AC"; std::wofstream ofs(L"c:\\temp\\bla.xml"); boost::archive::xml_woarchive oa(ofs); oa << boost::serialization::make_nvp("tag", wstr); } However in 1.60 it gives an exception, since it cannot convert the uniocde character. It seems that in 1.57 in xml_woarchive_impl the utf8_codecvt_facet is set, but this has been disabled in 1.60. What is the solution / workaround?

2 3

boost::property_tree::read_json fails on previously supported comments 1.59
by Millman, Mark 18 Dec '15

18 Dec '15

I recently upgraded from 1.56 to 1.59 and my hundreds of well commented JSON parameter files now all fail on the first commented line. This is true for /* */ and // style comments. The exception.what() string thrown by boost::property_tree::read_json(filename,propertyTree) is filename.json(linenumber): expected key string. As I remove comments from my test JSON file the line number advances to the next comment. Any hints as to how to deal with this? Removing all my comments is not an option. Thanks, Mark --

2 5

[boost.users] [statechart] question about event processing rules in orthogonal states.
by Ernesto Rodriguez Reina 18 Dec '15

18 Dec '15

Hi everyone, The statechart search for an event handler in post-order (children first then parent), however when we have orthogonal states involved then the behaviour I am experiencing is different. Let say we have an state X with two children [orthogonal] states Y and W. Assuming none of the handlers discard (or defer) an event, the handler for that event is called in the following order: Y, X, W, X. Note that the handler of X is called twice. The following code demonstrates the issue: <code> struct EvTest : sc::event< EvTest > {}; struct Test; struct Inner1; struct Inner2; struct Machine : sc::state_machine< Machine, Test > {}; struct Test : sc::simple_state<Test, Machine, mpl::list<Inner1, Inner2> >{ typedef sc::custom_reaction< EvTest > reactions; sc::result react( const EvTest){ std::cout << "Test" << std::endl; // return discard_event(); return forward_event(); } }; struct Inner1 : sc::simple_state<Inner1, Test::orthogonal< 0 > >{}; struct Inner2 : sc::simple_state<Inner2, Test::orthogonal< 1 > >{ typedef sc::custom_reaction< EvTest > reactions; sc::result react( const EvTest){ std::cout << "Inner2" << std::endl; return forward_event(); } }; </code> When the event "EvTest" is received, the output I am getting is: <output> Test Inner2 Test </output> If there something that I am missing? Best, Ernesto Rodriguez Reina

1 0

how to help an object to delete itself the boost way
by Ireneusz Szcześniak 18 Dec '15

18 Dec '15

Hi, In my code there are objects created, which delete themselves. They can do it in the constructor, or later in some function. This is ugly, and I'm getting some wierd behaviour, probably because of this bad design. Is there a good Boost way of resolving my problem? Thanks, Irek

6 10

Negative effect of expected-to-fail tests in compiler development projects.
by Sergey Sprogis 18 Dec '15

18 Dec '15

I’m wondering how difficult it would be to add into bjam one option, something like: -negative_tests=off which will not launch any of negative (expected to fail) tests. For people who use Boost to test newly developed compilers negative tests are quite a nuisance. I mean those tests which got “compile_fail” and “run_fail” status inside final *.xml file generated at the end of Boost regression testing. [ they are also marked similarly inside Jamfiles ] That’s because newly developed compilers in their early stages of implementation normally have bugs which produce a lot o false error messages for correct Boost code. Important task here is to extract those messages, to evaluate them, and to prove that they are indeed correct, but compiler is wrong. And when hundreds of such false error messages are mixing together with thousands of legitimate error messages produced by negative tests (there are > 700 of them in boost.1.59.0, for example) it becomes a non-trivial task to filter them out. So the natural desire is not to launch such tests at all. Another, slightly unpleasant effect of negative tests is that they appear during so called “test pass rate” calculations during compiler testing process. Typically managers responsible for compiler development want to know the progress in terms of what “test pass rate” produces new compiler for the whole Boost testing or for the testing of some specific libraries. Normally such pass rate is calculated as the ratio between number of passed tests and the total number of tests. But ideally, tests in those calculations should be correct, so if they fail, it 100% means compiler bugs. And here again, negative tests are not useful, and should be avoided to make calculations more accurate. On a side note, I think it could be also useful to add such total testing pass rates into the Boost Regression Dashboard, so the quality of every tested compiler could be easy to seen. Much more people could be interested in looking at such dashboard, I guess.

5 6

boost::mutex::scoped_lock crashes on OSX
by David Medine 17 Dec '15

17 Dec '15

I am using boost in a python extension and it is having some problems on OSX. Specifically, I use a mutex of type boost::mutex in both the constructor and destructor methods. The constructor goes fine and the destructor crashes. The code works perfectly well on both Windows and Linux. When my class is building itself, it calls functions like this one a bunch of times: void enter_headers_phase(bool phase_locked) { if (phase_locked) { boost::mutex::scoped_lock lock(phase_mut_); headers_to_finish_++; } } where phase_mut_ is of type boost::mutex. When debugging in XCode, I see nice looking members for phase_mut_ at this point in the program: phase_mut_ (boost::mutex) m (pthread_mutex_t) __sig (long) 1297437784 1297437784 __opaque char [56] "" Later, when I am destroying, I make a nearly identical call as before: ... { boost::mutex::scoped_lock lock(phase_mut_); shutdown_ = true; } ... but now, my phase_mut_ looks like it has an apparently empty _sig member: phase_mut_ (boost::mutex) m (pthread_mutex_t) __sig (long) __opaque (char [56]) "" When we go down the line of the mutex code in boost and finally arrive at line 62 in boost/thread/pthread/mutex.hpp: ret = ::pthread_mutex_lock(m); I get a segmentation fault: EXC_BAD_ACCESS It seems like this is becausem->_sig is now empty for some reason, but I can not determine why (XCode's debugger won't let me 'watch' this variable) but it should certainly be the case that it doesn't get touched between the apparently successful constructor routines and now. This all happens on OSX Yosemite. I am compiling my extension with LLVM6.1 and I have gnu++0x as the C++ language dialect and libc++ as the stdlib. I am linking statically to boost 1.57 which I built using this command: sudo ./b2 -a cxxflags="-stdlib=libc++" linkflags="-stdlib=libc++" toolset=clang link=static link=shared threading=multi install --prefix=/opt/local --layout-tagged THX! -David

2 4

HP fortify Scan Issues
by Ganesh 17 Dec '15

17 Dec '15

I have run HP fortify scan on our code which uses Boos libraries for checking security vulnerabilities. During the scan I got issues reported in below two files. 1. compressed_pair.hpp2. next_prior.hpp The issue reported as below.1. compressed_pair.hpp (In line 154) LINE 153: compressed_pair_imp(first_param_type x, second_param_type y)LINE 154: : first_type(x), second_(y) {} Abstract: The program reads data from just outside the bounds of allocated memory. Explanation: Buffer overflow is probably the best known form of software security vulnerability. Most software developers know what a buffer overflow vulnerability is, but buffer overflow attacks against both legacy and newly-developed applications are still quite common. Part of the problem is due to the wide variety of ways buffer overflows can occur, and part is due to the error-prone techniques often used to prevent them. In a classic buffer overflow exploit, the attacker sends data to a program, which it stores in an undersized stack buffer. The result is that information on the call stack is overwritten, including the function's return pointer. The data sets the value of the return pointer so that when the function returns, it transfers control to malicious code contained in the attacker's data. 2. next_prior.hpp (In line 30) LINE 29: template <class T>LINE 30 inline T next(T x) { return ++x; } Abstract: The function next() in next_prior.hpp reads data from just outside the bounds of ~tp~1 on line 30. Explanation: Buffer overflow is probably the best known form of software security vulnerability. Most software developers know what a buffer overflow vulnerability is, but buffer overflow attacks against both legacy and newly-developed applications are still quite common. Part of the problem is due to the wide variety of ways buffer overflows can occur, and part is due to the error-prone techniques often used to prevent them. In a classic buffer overflow exploit, the attacker sends data to a program, which it stores in an undersized stack buffer. The result is that information on the call stack is overwritten, including the function's return pointer. The data sets the value of the return pointer so that when the function returns, it transfers control to malicious code contained in the attacker's data. Question is: Is this really an issue with the code? or boost has taken care of this out of bound in its code?

2 1

Size(Count of elements) in boost::multi_index_container?
by Amit Prakash Ambasta 17 Dec '15

17 Dec '15

Hi, How do I find out the size or count of elements added to a multi_index_container?

2 1

Re: [Boost-users] [Boost-announce] Boost 1.60.0 rc1 is now available
by Jan Mikkelsen 17 Dec '15

17 Dec '15

Hi, (Resending to avoid bouncing on the lists.) I can build the libraries on FreeBSD 10.2 with clang 3.7 and gcc 5.2 in various modes. However, with both clang 3.7 and gcc 5.2 the following program fails to compile: #define BOOST_MOVE_USE_STANDARD_LIBRARY_MOVE #include <boost/thread.hpp> int main() { } This works fine with the same compilers and boost 1.59. The difference seems to be caused a change to boost/thread/detail/thread_group.hpp to use boost::csbl::unique_ptr<> instead of std::auto_ptr<>. Diff showing the change between 1.59 and 1.60 in thread_group.hpp and the compiler error messages below. Regards, Jan Mikkelsen $ diff -u thread/detail/thread_group.hpp ../../boost-1_60/boost/thread/detail/thread_group.hpp --- thread/detail/thread_group.hpp 2015-12-17 01:08:28.000000000 +1100 +++ ../../boost-1_60/boost/thread/detail/thread_group.hpp 2015-12-17 01:20:56.000000000 +1100 @@ -6,6 +6,7 @@ // (C) Copyright 2007-9 Anthony Williams #include <list> +#include <boost/thread/csbl/memory/unique_ptr.hpp> #include <boost/thread/shared_mutex.hpp> #include <boost/thread/mutex.hpp> #include <boost/thread/lock_guard.hpp> @@ -75,7 +76,7 @@ thread* create_thread(F threadfunc) { boost::lock_guard<shared_mutex> guard(m); - std::auto_ptr<thread> new_thread(new thread(threadfunc)); + boost::csbl::unique_ptr<thread> new_thread(new thread(threadfunc)); threads.push_back(new_thread.get()); return new_thread.release(); } $ g++5 -std=c++14 -isystem /usr/local/include/boost-1_60 twio/test_thread.cpp In file included from /usr/local/include/boost-1_60/boost/thread/csbl/memory/unique_ptr.hpp:16:0, from /usr/local/include/boost-1_60/boost/thread/detail/thread_group.hpp:9, from /usr/local/include/boost-1_60/boost/thread/thread.hpp:13, from /usr/local/include/boost-1_60/boost/thread.hpp:13, from twio/test_thread.cpp:2: /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp: In constructor 'boost::movelib::unique_ptr<T, D>::unique_ptr(boost::movelib::unique_ptr<T, D>&&)': /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp:530:29: error: 'move_if_not_lvalue_reference' is not a member of 'boost' : m_data(u.release(), ::boost::move_if_not_lvalue_reference<D>(u.get_deleter())) ^ /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp:530:68: error: expected primary-expression before '>' token : m_data(u.release(), ::boost::move_if_not_lvalue_reference<D>(u.get_deleter())) ^ In file included from /usr/local/include/boost-1_60/boost/thread/csbl/memory/unique_ptr.hpp:16:0, from /usr/local/include/boost-1_60/boost/thread/detail/thread_group.hpp:9, from /usr/local/include/boost-1_60/boost/thread/thread.hpp:13, from /usr/local/include/boost-1_60/boost/thread.hpp:13, from twio/test_thread.cpp:2: /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp: In constructor 'boost::movelib::unique_ptr<T, D>::unique_ptr(boost::movelib::unique_ptr<Y, D>&&, typename boost::move_upd::enable_up_moveconv_constr<T, D, U, E>::type*)': /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp:552:29: error: 'move_if_not_lvalue_reference' is not a member of 'boost' : m_data(u.release(), ::boost::move_if_not_lvalue_reference<E>(u.get_deleter())) ^ /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp:552:68: error: expected primary-expression before '>' token : m_data(u.release(), ::boost::move_if_not_lvalue_reference<E>(u.get_deleter())) ^ /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp: In member function 'boost::movelib::unique_ptr<T, D>& boost::movelib::unique_ptr<T, D>::operator=(boost::movelib::unique_ptr<T, D>&&)': /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp:581:26: error: 'move_if_not_lvalue_reference' is not a member of 'boost' m_data.deleter() = ::boost::move_if_not_lvalue_reference<D>(u.get_deleter()); ^ /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp:581:65: error: expected primary-expression before '>' token m_data.deleter() = ::boost::move_if_not_lvalue_reference<D>(u.get_deleter()); ^ /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp: In member function 'typename boost::move_upd::enable_up_moveconv_assign<T, D, U, E, boost::movelib::unique_ptr<T, D>&>::type boost::movelib::unique_ptr<T, D>::operator=(boost::movelib::unique_ptr<Y, D>&&)': /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp:603:26: error: 'move_if_not_lvalue_reference' is not a member of 'boost' m_data.deleter() = ::boost::move_if_not_lvalue_reference<E>(u.get_deleter()); ^ /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp:603:65: error: expected primary-expression before '>' token m_data.deleter() = ::boost::move_if_not_lvalue_reference<E>(u.get_deleter()); ^ $ clang++37 -std=c++14 -isystem /usr/local/include/boost-1_60 twio/test_thread.cpp In file included from twio/test_thread.cpp:2: In file included from /usr/local/include/boost-1_60/boost/thread.hpp:13: In file included from /usr/local/include/boost-1_60/boost/thread/thread.hpp:13: In file included from /usr/local/include/boost-1_60/boost/thread/detail/thread_group.hpp:9: In file included from /usr/local/include/boost-1_60/boost/thread/csbl/memory/unique_ptr.hpp:16: /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp:530:67: error: 'D' does not refer to a value : m_data(u.release(), ::boost::move_if_not_lvalue_reference<D>(u.g... ^ /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp:353:26: note: declared here template <class T, class D = default_delete<T> > ^ /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp:530:38: error: no member named 'move_if_not_lvalue_reference' in namespace 'boost' : m_data(u.release(), ::boost::move_if_not_lvalue_reference<D>(u.g... ~~~~~~~~~^ /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp:552:67: error: 'E' does not refer to a value : m_data(u.release(), ::boost::move_if_not_lvalue_reference<E>(u.g... ^ /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp:548:29: note: declared here template <class U, class E> ^ /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp:552:38: error: no member named 'move_if_not_lvalue_reference' in namespace 'boost' : m_data(u.release(), ::boost::move_if_not_lvalue_reference<E>(u.g... ~~~~~~~~~^ /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp:581:64: error: 'D' does not refer to a value m_data.deleter() = ::boost::move_if_not_lvalue_reference<D>(u.get_... ^ /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp:353:26: note: declared here template <class T, class D = default_delete<T> > ^ /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp:581:35: error: no member named 'move_if_not_lvalue_reference' in namespace 'boost' m_data.deleter() = ::boost::move_if_not_lvalue_reference<D>(u.get_... ~~~~~~~~~^ /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp:603:64: error: 'E' does not refer to a value m_data.deleter() = ::boost::move_if_not_lvalue_reference<E>(u.get_... ^ /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp:597:29: note: declared here template <class U, class E> ^ /usr/local/include/boost-1_60/boost/move/unique_ptr.hpp:603:35: error: no member named 'move_if_not_lvalue_reference' in namespace 'boost' m_data.deleter() = ::boost::move_if_not_lvalue_reference<E>(u.get_... ~~~~~~~~~^ 8 errors generated. > On 15 Dec 2015, at 02:54, Marshall Clow <mclow.lists(a)gmail.com <mailto:mclow.lists@gmail.com>> wrote: > > Release candidate files for 1.60.0 are available at: http://boost.cowic.de/rc/ <http://boost.cowic.de/rc/> > > 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_60_0_rc1.7z) = 7ce7f5a4e396484da8da6b60d4ed7661 > MD5 (boost_1_60_0_rc1.tar.bz2) = 65a840e1a0b13a558ff19eeb2c4f0cbe > MD5 (boost_1_60_0_rc1.tar.gz) = 28f58b9a33469388302110562bdf6188 > MD5 (boost_1_60_0_rc1.zip) = 0cc5b9cf9ccdf26945b225c7338b4288 > > Thanks! > > -- The release managers > > P.S. Release notes are here: http://www.boost.org/users/history/version_1_60_0.html <http://www.boost.org/users/history/version_1_60_0.html> > > _______________________________________________ > Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost-announce <http://lists.boost.org/mailman/listinfo.cgi/boost-announce>

1 0

multi_index::composite_key with char[]
by leyiliu 16 Dec '15

16 Dec '15

Hi,All: When I use the multi_index::composite_key with char[],I meet a error,when find a element。It seems to the make_tuple can't support the char[].Because the char[] more faster the std::string。So our project choose char[] as the column instead of std::string.So How should I do -- View this message in context: http://boost.2283326.n4.nabble.com/multi-index-composite-key-with-char-tp46… Sent from the Boost - Users mailing list archive at Nabble.com.

3 7

Regarding thread_specific_ptr's non-moveability
by Daniel Hofmann 15 Dec '15

15 Dec '15

thread_specific_ptr is currently (Boost 1.59) not moveable, since it defines a user-declared destructor and in addition (privately) user-declares the copy special member functions. Is this decision by design, or simply not yet updated to move semantics? >From what I can tell, a move should be possible along the lines of: auto* p = other.release(); // without invoking other's cleanup fn reset(p); // potentially differente cleanup fn associated The only problematic situation I can think of is a scenario with two semantically different custom cleanup function where a move eventually triggers only the second one. But this is behavior is consistent with move semantic. Cheers, Daniel J H References: > http://www.boost.org/doc/libs/1_59_0/doc/html/thread/thread_local_storage.h… > http://www.boost.org/doc/libs/1_59_0/boost/thread/tss.hpp

2 1

[smart_ptr] BugTicket #11843, Hash not supported for shared_ptr<T[N]>
by Gagan Kumar 15 Dec '15

15 Dec '15

I have tried to solve problem for hash not supported for shared_ptr<T[N]> https://github.com/boostorg/smart_ptr/pull/22 However, I wanted to know Is there a way to solve hash problem with shared_ptr<T[]>. Since hash here depends on all element and we can't get array size in this case. Still, does there exists way which I may be missing.

1 0

[PBGL] Very slow graph creation
by Quaglino Alessio 15 Dec '15

15 Dec '15

Hello, I am trying to parallelize my code and while the rest was easy, I have spent more than 1 month on the BGL part (despite it being only 2% of the runtime) without any success. I have written the following code: using namespace boost; using boost::graph::distributed::mpi_process_group; typedef adjacency_list<vecS, distributedS<mpi_process_group, vecS>, undirectedS> Graph; typedef iterator_property_map<std::vector<int>::iterator, property_map<Graph, vertex_index_t>::type> LocalMap; int getConnectedComponents(std::vector< std::pair<int,int> > &arcs, const std::vector<int> &isActive) { std::vector<int> localComponent(nV+1); std::vector<int> globalComponent(nV+1); Graph G(nV+1); // VERY SLOW!! // Add edges to the graph VERY SLOW!! for(std::vector<std::pair<int,int> >::iterator it = arcs.begin(); it != arcs.end(); ++it) if( isActive[arcCnt++]==0 ) add_edge(vertex(it->first,G),vertex(it->second,G),G); synchronize(G); // Compute number of connected components VERY SLOW with -np 2 LocalMap components(localComponent.begin(),get(vertex_index, G)); num = connected_components_ps(G, components); // Collect the global component vector components.set_max_ghost_cells(0); for( int i = 0; i < nV+1; i++ ) globalComponent[i] = get(components, vertex(i, G)); synchronize(components); } That parallelizes the following implementation: typedef adjacency_list <vecS, vecS, undirectedS> SerialGraph; int getConnectedComponents(std::vector< std::pair<int,int> > &arcs, const std::vector<int> &isActive) { int rank; std::vector<int> globalComponent(nV+1); SerialGraph G(nV+1); MPI_Comm_rank(PETSC_COMM_WORLD, &rank); // Only rank 0 has got the arc data structure if( rank == 0 ) { // Add edges to the graph for(std::vector<std::pair<int,int> >::iterator it = arcs.begin(); it != arcs.end(); ++it) if( isActive[arcCnt++]==0 ) boost::add_edge(it->first,it->second,G); num = connected_components(G, &globalComponent[0]); } // Propagate size for allocation MPI_Bcast( &num, 1, MPI_INT, 0, PETSC_COMM_WORLD ); // Propagate connected components MPI_Bcast( &globalComponent[0], nV+1, MPI_INT, 0, PETSC_COMM_WORLD ); However, the parallel version on a single core is about 5 times slower than the non-parallel one. Moreover, on two cores it becomes 15-20 times slower. What am I doing wrong? Regards, Alessio

1 0