February 2018 - Boost-users - lists.preview.boost.org

[interprocess] Determine size of managed_shared_memory
by Kai Benndorf 12 Oct '19

12 Oct '19

Hi all, I want to create a managed_shared_memory in exactly the size I needed. Therefore I do something like this: size = sizeof( common_managed_shared_memory ); size += sizeof( MyClassInSharedMem ); managed_shared_memory segment(create_only, "name", size ); MyClassInSharedMem allocates itself some memory by segment->allocate(). I added this also to the size, but get a boost::interprocess::bad_alloc exception on allocating this additional memory. Are there some magic bytes I forgot to add? Many thanks in advance Kai

3 2

boost::filtered_graph filtering on vertex color... if I only had a colormap
by Brian Davis 19 Jul '19

19 Jul '19

Lets say I have a graph g (of type graph_t) and a color map (boy I wish I had a colormap): typedef boost::property_map<graph_t, boost::vertex_color_t>::type color_map_t; or is it? auto indexmap = boost::get(boost::vertex_index, g); auto colors = boost::make_vector_property_map<boost::default_color_type>(indexmap); or maybe? auto indexmap = boost::get(boost::vertex_index, g); auto colors = boost::make_vector_property_map<boost::vertex_color_t>(indexmap); or possibly some goop found on the internet? boost::make_iterator_property_map(color_map.begin(), boost::get(boost::vertex_index, data_->g), color_map[0]) or other internet goop? typedef boost::property_map<GraphType, boost::vertex_color_t>::type color_map_t; color_map_t colorMap; boost::color_map(colorMap)) or other internet goop?? std::vector<int> colorMap(num_vertices(g), boost::white_color); so confused... (certainly not surprised why bgl is not a part of stl) Can someone help explain the various types and ways of specifying colormaps and the logic for each way and means? and a function (if I could get that far without VS syntax highlighter going bonkers and gobs of compiler errors.. you know the type): make_pretty(g, colormap ) which colors the graph colormap and a filter struct for boost::filtered_graph<Graph, EdgePredicate, VertexPredicate> (where I wonder why if I only want to filter vertices I have to set boost::keep_all for EdgePredicate... sigh... I know the answer... and also know how long it found me to find/figure out boost::keep_all - and if you are looking for boost::keep_all and found it here... your welcome ;-) ) and modifying (probably incorrectly) boost::filtered_graph example to: template <typename ColorMap> struct vertex_color_filter_eq { vertex_color_filter_eq() { } vertex_color_filter_eq(ColorMap colormap/*, boost::default_color_type c*/) : m_colormap(colormap)/*, m_c(c)*/ { } template <typename Vertex> bool operator()(const Vertex& v) const { return 0 < get(m_colormap, v); //return c < get(m_colormap, v); } ColorMap m_colormap; //boost::default_color_type m_c; }; auto indexmap = boost::get(boost::vertex_index, g); auto colors = boost::make_vector_property_map<boost::default_color_type>(indexmap); // some where here I would call make_pretty(g, colors) typedef property_map<graph_t, boost::vertex_color_t>::type ColorMap_t; vertex_color_filter<ColorMap_t> filter(colors); // boost::filtered_graph<graph_t, boost::keep_all, VertexPredicate> boost::filtered_graph<graph_t, boost::keep_all, vertex_color_filter> fg(g, filter); // Later say if I could get it compile and better if it compiled and worked I could move on and auto color = boost::red_color; typedef property_map<graph_t, boost::vertex_color_t>::type ColorMap_t; vertex_color_filter<ColorMap_t> filter(colors, color); boost::filtered_graph<graph_t, boost::keep_all, vertex_color_filter> fg(g, filter); Maybe if I color them yellow, I can then reach the wizard pull back the curtain and ask how to create a color map... If I only had a color map.

4 6

[Test] Test executable crashes always
by Nate Widmyer 26 Jun '18

26 Jun '18

Hello, I've made some unit tests with boost test before without issue, but on my latest rather ordinary test, I've been able to get Boost test to crash the executable in debug and release mode when boost test is making a report before terminating. test_exe.exe!boost::unit_test::framework::get(unsigned long id, boost::unit_test::test_unit_type t) Line 1388 C++ test_exe.exe!boost::unit_test::framework::get<boost::unit_test::test_unit>(unsigned long id) Line 208 C++ test_exe.exe!boost::unit_test::results_reporter::make_report(boost::unit_test::report_level l, unsigned long id) Line 175 C++ test_exe.exe!boost::unit_test::unit_test_main(boost::unit_test::test_suite *(*)(int, char * *) init_func, int argc, char * * argv) Line 235 C++ test_exe.exe!main(int argc, char * * argv) Line 287 C++ That's the default report_level of 'confirm' If I use report_level=short or detailed, I get a similar crash: test_exe.exe!boost::unit_test::framework::get(unsigned long id, boost::unit_test::test_unit_type t) Line 1388 C++ test_exe.exe!boost::unit_test::framework::get<boost::unit_test::test_suite>(unsigned long id) Line 207 C++ test_exe.exe!boost::unit_test::traverse_test_tree(unsigned long id, boost::unit_test::test_tree_visitor & V, bool ignore_status) Line 365 C++ test_exe.exe!boost::unit_test::results_reporter::make_report(boost::unit_test::report_level l, unsigned long id) Line 185 C++ test_exe.exe!boost::unit_test::unit_test_main(boost::unit_test::test_suite *(*)(int, char * *) init_func, int argc, char * * argv) Line 235 C++ test_exe.exe!main(int argc, char * * argv) Line 287 C++ Here's the get() method: test_unit& get( test_unit_id id, test_unit_type t ) { test_unit* res = impl::s_frk_state().m_test_units[id]; BOOST_TEST_I_ASSRT( (res->p_type & t) != 0, internal_error( "Invalid test unit type" ) ); return *res; } Crash occurs on line 2 of the function. I logged 'res' each time the function is called and many times its valid, but when it runs this one test (by itself or with others), 'res' always ends up being NULL. id seems to always be '1', and t is TUT_ANY (17) if generating a confirmation report or TUT_SUITE (16) if generating a short or detailed report. If I use the option '--report_level=no', then execution does complete fine. I followed through on some process of elimination and it seems to be related to using a fixture? Having no test body still causes the crash. It wasn't until I removed the fixture (and its usage in the body of the test) did the executable not crash. I ensured that I was using the standard fixture model found here: http://www.boost.org/doc/libs/1_64_0/libs/test/doc/html/boost_test/tests_or… OS is Win 7 w/ VS2015 boost is 1.64.0 and was built with: b2 toolset=msvc-14.0 link=static runtime-link=static variant=debug address-model=64 --without-atomic --without-container --without-context --without-coroutine --without-coroutine2 --without-graph --without-graph_parallel --without-iostreams --without-locale --without-math --without-mpi --without-serialization --without-signals --without-type_erasure --without-wave -j4 --abbreviate-paths --without-python -d2 Any help would of something to look at or change would be appreciated. Thanks for your time, Nathan

2 4

Pointer Container Library changes in 1.66.0
by Kevin Hardman 06 Apr '18

06 Apr '18

I am seeing changes in the Pointer Container Library that are preventing my existing application from compile. I see no notice of changes in the 1.66.0 release notes <http://www.boost.org/users/history/version_1_66_0.html <http://www.boost.org/users/history/version_1_66_0.html>>. Is there documentation on the changes and what is required to migrate existing users to the new library? —kevin

5 10

[fiber] Fiber context switching during stack unwinding...
by Oliver Seiler 01 Mar '18

01 Mar '18

(Hopefully I'm not writing an essay for something that is a gcc/libstdc++/Linux bug) I've been using Boost.Context for a number of years now, to do something similar to what Boost.Fiber does. We recently upgraded our development target environment to Debian 9 (gcc 6.3), and moved up to Boost 1.62.0, so I've been looking at Boost.Fiber to replace how we're using Boost.Context (because it looks like it could greatly simplify our code, among other things). One thing I ran into with Boost.Context was that it interacted in some interesting ways with exception handling, and I haven't really seen any discussion about it. Specifically, switching contexts with an active exception (so during stack unwinding, or in a catch block) would result in weird behaviour in the other context (std::uncaught_exception() returning true, std::current_exception() returning the exception from the other context, try/catches in the new context causing new problems when switching back to the original context which is in the middle of stack unwinding, etc). I chalked that up to limitations of the context switching, though I haven't seen any mention of avoiding context switches with active exceptions; presumably this is some interaction with gcc's exception implementation, and would maybe need support from libunwind to save/restore the exception state (I don't recall if I checked if clang produced similar results, though I'll probably give it a go when I have the chance). So I wasn't surprised when I ran into similar problems with Boost.Fiber. I am wondering why this aspect of usage isn't discussed in any of the documentation; it would be nice if these libraries could work well with language features like exceptions, but I understand that the language spec doesn't really speak to the stack switching happening under-the-hood. It actually wouldn't surprise me that this wasn't even a problem in all environments, and just happens to be an artifact of the Linux implementation, but then we're either talking about unspecified or undefined behaviour, but would be nice to know. Even in the documents going through the C++ standards committee I don't see much mention of how things like coroutines should interact with something like std::current_exception() (perhaps because it is assumed to "just work" and doesn't need additional comment?) Anyone else ever run into issues related to this? Seems like any use of RAII for exception handling would easily run into problems. Mostly I've just found ways to work around it (e.g., not allowing the context switch if std::uncaught_exception() or std::current_exception() indicate it would be unsafe to do so, typically by throwing another exception) but this tends to make libraries harder to use correctly (e.g., can't really allow context switching in a destructor, or in a function called from a destructor, etc). Cheers Oliver

3 4

PPA for boost
by legalize+jeeves＠mail.xmission.com 28 Feb '18

28 Feb '18

[Please do not mail me a copy of your followup] Hello fellow boosters, I am one of the maintainers of the C++ language track on exercism.io. When I first implemented this language track, we didn't have good regex support in g++'s standard library yet, so we needed boost.regex to fill that gap. Since we were already going to use boost.regex, I used boost.test for the unit test support. At the time, I found a PPA that was providing prebuilt binaries for ubuntu. Since that time, the need for boost.regex has gone away because the standard library implementation has been completed for g++. Along the way, the biggest source of problems in getting people going with C++ has been boost. For Windows users, it's annoying that CMake won't "just find" boost when you install it because there's no uniform mechanism between boost, packaging and CMake to say where boost got installed. For Mac users, they end up having difficulty with homebrew and fighting with Xcode. For Linux users, the PPA has gone stale and they end up being forced to build from scratch and get sent off into a ditch. In the meantime, we've got conan at stable 1.0 version right now, but that introduces a python dependency and then a conan dependency and for beginners I don't know if that's a path less fraught with peril. We've done some work on a branch to ditch boost.test and therefore boost entirely and switch the unit test framework to Catch. We have yet to "drop the hammer" on that and bring it onto master because reasons. (I am not the developer of that branch, but I suspect it hasn't happened yet because of the other integration issues that are needed; namely a bunch of documentation has to be updated and there will be teething pains with anyone who has already downloaded existing exercises and written them against boost.test.) Now, I find it hard to believe that I am the only one that has suffered from these issues all this time. Sure, if you only use header-only libraries then you don't need to build anything, but there are more than a few libraries in boost that either require or benefit significantly from being compiled. In all this time, have we not managed to come up with a stable, reliable PPA archive with ongoing support for boost for linux users? All I seem able to find are half-heartedly maintained PPAs with stale versions of boost on them. I have even tried to navigate going about setting up my own PPA in order to solve the problem myself, but I found launchpad.net and the whole PPA process so poorly documented and explained that I couldn't make head-nor-tails of it and gave up. (And this is coming from someone who has done more than my fair share of packaging and installer work on multiple platforms.) -- "The Direct3D Graphics Pipeline" free book <http://tinyurl.com/d3d-pipeline> The Terminals Wiki <http://terminals-wiki.org> The Computer Graphics Museum <http://computergraphicsmuseum.org> Legalize Adulthood! (my blog) <http://legalizeadulthood.wordpress.com>

2 1

Missing libs
by Tim Burgess 28 Feb '18

28 Feb '18

Hi, I downloaded the modular Boost master from Git Hub, as per instructions and built it on Windows as follows: .\Bootstrap .\b2 headers .\b2 This generated (amongst others) the following libs for boost::filesystem: libboost_filesystem-vc141-mt-gd-x32-1_67.lib libboost_filesystem-vc141-mt-gd-x64-1_67.lib libboost_filesystem-vc141-mt-x32-1_67.lib libboost_filesystem-vc141-mt-x64-1_67.lib I then tried to build a static release X64 version of my application using VS2017 and got: 1>LINK : fatal error LNK1104: cannot open file 'libboost_filesystem-vc141-mt-s-x64-1_67.lib' Obviously, I don't have this file, though I had thought that .\b2 was supposed to build all targets. So, I tried: .\b2 link=static This still does not generate the required file - where am I going wrong, please? BTW, I have checked that my project's library path points to the appropriate boost\stage\lib folder. Best wishes. Tim Burgess

2 2

[Thread] Version 1.66.0 incompatible with ISOLATION_AWARE_ENABLED
by Jim Wilburn 28 Feb '18

28 Feb '18

I work on a Windows code base that I have been updating from Boost 1.64.0 to Boost 1.66.0. One of the projects sets ISOLATION_AWARE_ENABLED to 1 (refer to https://msdn.microsoft.com/en-us/library/windows/desktop/bb773175(v=vs.85).…). #define ISOLATION_AWARE_ENABLED 1 The project also includes <boost/thread.hpp>. The Windows SDK version is 10.0.16299.0. The include chain is: <boost/thread.hpp> <boost/thread/thread.hpp> <boost/thread/thread_only.hpp> <boost/thread/win32/thread_data.hpp> <boost/thread/win32/thread_primitives.hpp> <boost/detail/winapi/dll.hpp> <boost/winapi/dll.hpp> The last include contains constructions such as: BOOST_SYMBOL_IMPORT boost::winapi::HMODULE_ WINAPI LoadLibraryA(boost::winapi::LPCSTR_ lpFileName); With a nonzero ISOLATION_AWARE_ENABLED, this results in redefinition errors, e.g.: .../boost/winapi/dll.hpp(38): error C2375: 'IsolationAwareLoadLibraryA': redefinition; different linkage ...\include\10.0.16299.0\um\winbase.inl(287): note: see declaration of 'IsolationAwareLoadLibraryA' Jim Wilburn

2 1

all roots of a polynomial
by mdebonis 22 Feb '18

22 Feb '18

Hello, I found Boost while searching for a C++ library which can do accurate polynomial root finding. The problem is that the documentation does not make it very clear how to implement this tool. Does anyone by any chance have some sample code on implementing this feature or steer me in the direction of an example in the documentation? I would like to use the best decimal place accuracy (it appears that you can get 50-place here?) Thanks! -- Sent from: http://boost.2283326.n4.nabble.com/Boost-Users-f2553780.html

4 3

Boost improperly including NTVERP.H
by Brian Catlin 22 Feb '18

22 Feb '18

Boost 1.66.0 is causing the Windows file NTVERP.H to be included (defining macros such as VER_DEBUG, VER_PRODUCTVERSION_STR, etc), preventing the definition of those macros by the application. This behavior didn't exist in 1.65.1, and breaks a standard Windows paradigm (it is very common for a Windows program to have a Version.h file containing the version resource macros (that are also defined in NTVERP.H), and then include Version.h in the main CPP file and the Resources.RC file). The NTVERP.H file contains the version information for the Windows SDK, and should NOT be included in an application! NTVERPH.H is included in: boost.1.66.0.0\lib\native\include\boost\predef\platform\windows_uwp.h boost.1.66.0.0\lib\native\include\boost\winapi\config.hpp

1 0