- Boost-users - lists.boost.org

Memory header not found
by Sudhanshu Dubey 11 Sep '18

11 Sep '18

Hey guys, I am new here and wanted some help from you guys. I was trying to upgrade and compile the FElt software, a software for finite element analysis. The source code is here : https://github.com/AkMecha/felt Now, the problem is that during compilation, one of the source file is including some header file from the boost library. The chain of including goes on and reaches the memory.hpp file which is present in /usr/include/boost/config/no_tr1/ directory. This memory.hpp is in turn including the memory header file from the standard C++ library, but it can't find that memory header file. And this is strange because I personally went and checked that the memory header is there in the standard location, as you can see below : sudhanshu@sudhanshu-Inspiron-15-3567:/usr/include/c++/7.3.0$ ls algorithm cerrno complex cstdio decimal initializer_list map queue stdlib.h typeindex any cfenv complex.h cstdlib deque iomanip math.h random streambuf typeinfo array cfloat condition_variable cstring exception ios memory ratio string type_traits atomic chrono csetjmp ctgmath experimental iosfwd mutex regex string_view unordered_map backward cinttypes csignal ctime ext iostream new scoped_allocator system_error unordered_set bits ciso646 cstdalign cuchar fenv.h istream numeric set tgmath.h utility bitset climits cstdarg cwchar forward_list iterator optional shared_mutex thread valarray cassert clocale cstdbool cwctype fstream limits ostream sstream tr1 variant ccomplex cmath cstddef cxxabi.h functional list parallel stack tr2 vector cctype codecvt cstdint debug future locale profile stdexcept tuple The error message shown during the compilation is : [ 51%] Building C object src/Burluap/CMakeFiles/burluap.dir/apply.c.o In file included from /usr/include/boost/smart_ptr/shared_ptr.hpp:23:0, from /usr/include/boost/shared_ptr.hpp:17, from /home/sudhanshu/Pictures/felt/include/matrix.h:23, from /home/sudhanshu/Pictures/felt/src/Burluap/descriptor.h:35, from /home/sudhanshu/Pictures/felt/src/Burluap/trap.h:31, from /home/sudhanshu/Pictures/felt/src/Burluap/apply.c:29: /usr/include/boost/config/no_tr1/memory.hpp:21:12: fatal error: memory: No such file or directory # include <memory> ^~~~~~~~ compilation terminated. src/Burluap/CMakeFiles/burluap.dir/build.make:62: recipe for target 'src/Burluap/CMakeFiles/burluap.dir/apply.c.o' failed make[2]: *** [src/Burluap/CMakeFiles/burluap.dir/apply.c.o] Error 1 CMakeFiles/Makefile2:504: recipe for target 'src/Burluap/CMakeFiles/burluap.dir/all' failed make[1]: *** [src/Burluap/CMakeFiles/burluap.dir/all] Error 2 Makefile:129: recipe for target 'all' failed make: *** [all] Error 2 So can I get some help as to what should I do to make the memory.hpp find the memory header file? -- Sudhanshu Dubey https://hacksd.wordpress.com https://github.com/Sudhanshu-Dubey14

1 0

ASIO?
by tim＠raisedbar.net 05 Sep '18

05 Sep '18

Hi, I'm looking for a mechanism allowing me to enumerate serial (Com) ports on Windows/Mac. I also need to be able to retrieve the name of a given serial device. I've had a quick look at boost::asio, but it isn't clear to me if it can do what I want. I'd be grateful for any advice, including alternatives if boost and/or ASIO aren't up to the job. Best wishes. Tim Burgess Raised Bar Ltd. E: <mailto:tim@raisedbar.net> tim(a)raisedbar.net M: +44 (0)7989 486976

3 2

mcgregor_common_subgraphs usage/complexity/performance/runtime
by Julian Wolf 01 Sep '18

01 Sep '18

Hello everyone, I'm new to Boost and today I tried to use the mcgregor_common_subgraphs functions but can't find an explanation why the function shows such bad performance when I use it. According to the documentation [1] the algorithm runs in O(V1 * V2) for the entire space with V being the number of vertices in the two graphs. So not great, but I'd expect that to run in reasonable time for graphs with less than 100 vertices. But in my case it doesn't, therefore any help appreciated whether I have some error or misconception on my side or whether this is expected performance. What I'm trying to achieve: I want to check whether two graphs have a common subgraph of size x. x depends on the size of smaller graph, I'm using a factor y for this with 0 < y < 1. With smaller values for x (e.g. x = 0.7 * V1) the example returns very fast, which makes sense since a smaller part of the search space needs to be explored. If I increase the factor to 0.8 * V1 or even run through the whole search space the program does not terminate within 10 minutes (where I canceled) I'll attach a working example as well as two graphs (44 and 46 vertices) showing the behaviour described. # gcc CommonSubgraph.cpp -o common_subgraph_example -lboost_graph -lstdc++ # time ./common_subgraph_example 0.7 graph1.dot graph2.dot Factor 0.7 Size: 30 real 0m0.043s # time ./common_subgraph_example 0.8 graph1.dot graph2.dot Factor 0.8 ^C real 10m39.741s My system runs openSUSE Tumbleweed with Boost 1.67 > rpm -q --whatprovides /usr/include/boost/graph/mcgregor_common_subgraphs.hpp libboost_headers1_67_0-devel-1.67.0-2.3.x86_64 Thank you! Julian [1] https://www.boost.org/doc/libs/1_68_0/libs/graph/doc/ mcgregor_common_subgraphs.html

2 2

Testing and faking ASIO I/O objects
by Cristian Morales Vega 31 Aug '18

31 Aug '18

How is people around here testing code which uses ASIO? I guess something like GIVEN("SOMETHING THAT STARTS BY DOING A NAME RESOLVER IS STARTED") fakeResolver.async_resolve("www.google.com", "80", HandlerCallingConnect); // This stores the completion handler (and an executor_work_guard) internally WHEN("THE RESOLVER RETURNS WITHOUT ERROR") fakeResolver.completeWithFakeResult("1.2.3.4", error_code()); // This post() the completion handler with the arguments you give it here io_context.poll(); THEN("A CONNECTION IS ATTEMPTED") EXPECT_CALL(fakeSocket, async_connect()); would be nice. But the "stores the completion handler" part doesn't look that simple. I have found some attempts online, but they seems to store the completion handlers into a vector of std::function and post them through std::bind, so possibly using the wrong executor, etc. There is any good set of ASIO I/O object fakes available?

2 2

linker errors with latest Visual Studio 15.8.2 update
by Damien 30 Aug '18

30 Aug '18

All, It looks like with 15.8.2 you might need to rebuild if you are linking to Boost compiled with earlier versions in the 15.6 or 15.7 series. Screens full of unresolved symbols..... Just a heads-up. Damien

1 0

[boost.test] parsing runtime arguments
by Olaf Peter 27 Aug '18

27 Aug '18

Hi, how get an 'user' API for the boost.test like this: testsuite_xyz -- --help or even testsuite_xyz -- --save-path=/foo to get the user aspects. I know I can get the args by int argc{ framework::master_test_suite().argc }; char** argv{ framework::master_test_suite().argv }; Further, at test/include/boost/test/utils/runtime/cla/parser.hpp are some utils used by boost.test self. Anyway, the problem rises to get the behaviour desribed above, simply using a global fixture with: fixture::fixture() { int argc{ utf::framework::master_test_suite().argc }; char** argv{ utf::framework::master_test_suite().argv }; for (int i = 0; i != argc; ++i) { std::cerr << "test_observer_fixture: " << argv[i] << "\n"; } } doesn't give the expected results - all the BOOST_AUTO_TEST_SUITE test cases are called even they depend on this arguments. Thanks, Olaf

2 3

Boost.Test BOOST_TEST/BOOST_REQUIRE_MESSAGE message arg problems
by Olaf Peter 27 Aug '18

27 Aug '18

Hello Boost Users, for my project I'm testing using BOOST_AUTO_TEST_SUITE and data driven test sets (BOOST_DATA_TEST_CASE). The goal is to check the proper working of a parser by checking success as such and compare the result with a "expected" data (string) vector. In case of parse error, I want to write the captured cout/cerr by boost.test into a file. In case of compare error the parsed result also shall be written to a file. The fragments show the concept/current state: namespace x3_test { // aka utils .... /* [Detect if boost test case failed]( * https://stackoverflow.com/questions/18466857/detect-if-boost-test-case-fail…) */ bool current_test_passing() { using namespace boost::unit_test; auto const id = framework::current_test_case().p_id; auto const test_results = results_collector.results(id); return test_results.passed(); } std::string report_diagnostic(fs::path const& test_case_name, std::string const& input, std::string const& result) { ... if(!x3_test::current_test_passing()) { // only write in case of failed test test_case_result_writer result_writer(test_case_name); result_writer.write(result); } return ss.str(); } } // x3_test ... BOOST_AUTO_TEST_SUITE(.... BOOST_DATA_TEST_CASE( declaration, dataset.input() ^ dataset.expect() ^ dataset.test_case_name(), code, expect_AST, test_case_name) { using attribute_type = ast::declaration; auto const parser = parser::declaration; x3_test::testing_parser<attribute_type> parse; auto [parse_ok, parsed_AST] = parse(code, parser); BOOST_TEST(parse_ok); BOOST_REQUIRE_MESSAGE(parse_ok, x3_test::report_diagnostic(test_case_name, code, parsed_AST) ); BOOST_TEST(parsed_AST == expect_AST, btt::per_element()); BOOST_REQUIRE_MESSAGE(x3_test::current_test_passing(), x3_test::report_diagnostic(test_case_name, code, parsed_AST) ); } BOOST_AUTO_TEST_SUITE_END() After some testing, the error count is twice the real (obviously here) but it's all above a work arround. It seems, that BOOST_REQUIRE_MESSAGE evaluates always the message predicate (shown by using --log_level=all) nevertheless the test case successed or failed. If failed it is streamed to the console. My first thoughs where that the message is called only in case of test failure which isn't true. The problem is now, that my report_diagnostic function write the test output (parsed_AST) to a special dir/file for later evaluating etc. The intuitive way would be: std::string report_diagnostic(fs::path const& test_case_name, std::string const& input, std::string const& result) { ... test_case_result_writer result_writer(test_case_name); result_writer.write(result); return ss.str(); } BOOST_REQUIRE_MESSAGE(parse_ok, // call diagnostic only in failure case and write then x3_test::report_diagnostic(test_case_name, code, parsed_AST) ); BOOST_REQUIRE_MESSAGE((parsed_AST == expect_AST, btt::per_element()), // call diagnostic only in failure case and write then x3_test::report_diagnostic(test_case_name, code, parsed_AST) ); Hence not depending on current_test_passing() any more. How to get this? Further, how can I get the captured std::cerr in case of '!parse_ok' inside report_diagnostic? Thanks, Olaf

2 10

[signals2] Warn unused result
by Deven 24 Aug '18

24 Aug '18

Hi, I am using boost::signals2 library for event callback mechanism. The event register callback function returns a boost::signals2::scope_connection object for optional disconnect by the function registering itself. I want to mandate using the returns result (scoped_connection) object by the user. So, I use the gcc attribute "__attribute__( ( warn_unused_result ) )" with the function. The problem is that the compiler does not warn for unused result. Any ideas, why would this happen? I am using Linux with g++ 4.8.4 and c++11 is enabled. Boost version: 1.67.0 I have attached a minimal working example. Thanks, -Deven

1 0

Setting cxxflags for intel compiler
by Michalke, Simon 23 Aug '18

23 Aug '18

Hi, in order to compile boost I need to tell the intel compiler which gcc version to use. I am trying to achive this with a user-config.jam using intel-linux : icpc : <cxxflags> -gcc-name=gcc-7 -gxx-name=g++-7 ; But none of the flags appear in the compile log. Which syntax am I supposed to use to get them working? Regards, Simon

3 2

[multi_index] Introducing C++17 terse key specification syntax
by Joaquin M López Muñoz 22 Aug '18

22 Aug '18

Following some ideas discussed about a year ago, I've added a new boost::multi_index::key construct allowing for a very terse specification of predefined key extractors in C++17 compliant environments. For instance, given struct phonebook_entry { std::string family_name; std::string given_name; std::string phone_number()const; }; int num_calls(const phonebook_entry&); a classical definition like typedef multi_index_container< phonebook_entry, indexed_by< ordered_unique< const_mem_fun<phonebook_entry,std::string,&phonebook_entry::phone_number> >, ordered_unique< global_fun<const phonebook_entry&,int,num_calls> >, ordered_non_unique< composite_key< phonebook_entry, member<phonebook_entry,std::string,&phonebook_entry::family_name>, member<phonebook_entry,std::string,&phonebook_entry::given_name> > > > > phonebook; can now be written as typedef multi_index_container< phonebook_entry, indexed_by< ordered_unique< key<&phonebook_entry::phone_number> >, ordered_unique< key<num_calls> >, ordered_non_unique< key<&phonebook_entry::family_name,&phonebook_entry::given_name> > > > phonebook; If you have a C++17 compiler, you can play with this new feature by simply copying https://github.com/boostorg/multi_index/blob/develop/include/boost/multi_in… into your local Boost installation. More info at https://www.boost.org/doc/libs/develop/libs/multi_index/doc/tutorial/key_ex… https://www.boost.org/doc/libs/develop/libs/multi_index/doc/reference/key_e… Comments and suggestions, specially before Boost 1.69 ships and this becomes more or less frozen, are most welcome. Best regards, Joaquín M López Muñoz

1 0