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

Conversion from exception_ptr to shared_ptr
by rajesh daggumati 19 Oct '21

19 Oct '21

HI, please help me to solve below issue. Shared pointers are used in my code. class Exception : public virtual std::exception,public virtual boost::exception { --- }; class ExceptionTestException : public Exception { --- }; boost::shared_ptr<Exception> exceptionptr_t; try { Boost_Throw_function(ExceptionTestException("Hiii")); } catch(...) { boost::exception_ptr exceptionptr = boost::current_exception(); try { boost::rethrow(exceptionptr) } catch(boost::exception &e) { //here i need to assign this 'e' to shared pointer but this e is not in heap so then how to assign this e to sharedptr(exceptionptr_t). 1 way is : boost::exception *ep = new ExceptionTestException(e); exceptionptr_t(ep); here we know that exception_ptr is having ExceptionTestException but in some cases, we dont know which type of excepiton we have in exception_ptr in runtime. then how to assign to shared_ptr

1 0

Version 77 download not a bz2 file
by Chris 16 Oct '21

16 Oct '21

Trying to use version 77 I get a "wrong checksum" message: Failed to download sources: module boost: Wrong sha256 checksum for boost_1_77_0.tar.bz2, expected "fc9f85fc030e233142908241af7a846e60630aa7388de9a5fafb1f3a26840854", was "4ffd1325229da00455ae95a68de976d053e46315628ffcd57a433238e74e7eab" If I change the checksum in the download command to that shown above I get a "not a .bz2 file" message. Chris

2 4

Thread crash in destructor
by Alexander Carôt 08 Oct '21

08 Oct '21

Hi all, I had been using an old boost system and boost thread lib for a couple of years without upgrading it since approx. 2014. With this old lib the following (simplified) code works fine on OSX: #include "test.h" test::test() { boost::thread::attributes attrs; /// START TRACEROUTE THREAD #if defined(BOOST_THREAD_PLATFORM_WIN32) res = SetThreadPriority(attrs.native_handle(), THREAD_PRIORITY_NORMAL); #elif defined(BOOST_THREAD_PLATFORM_PTHREAD) pthread_attr_setschedpolicy(attrs.native_handle(), SCHED_FIFO); #endif tracerouteThread = boost::thread( attrs, boost::bind(&test::performTraceroute, this)); } void test::performTraceroute() { cout << "START TRACEROUTE" << endl; string exec = ""; cout << "DO FURTHER STUFF BUT THIS IS JUST A DEMO ..." << endl; cout << "THREAD OVER" << endl; } Using the current boost 1_77 still does execute the thread but then it crashes within the destructor. Adding tracerouteThread.join() or .detach() after launching the thread does not help either. Can anyone help how to resolve this issue ? Thanks a lot in advance, best Alex -- http://www.carot.de Email : Alexander(a)Carot.de Tel.: +49 (0)177 5719797

4 5

Re: [Boost-users] Thread crash in destructor
by Alexander Carôt 08 Oct '21

08 Oct '21

see below – scanning my own info I realized: >>Thread 4 Crashed: >>0 libboost_thread.dylib 0x000000010e4a4570 tls_destructor + 96 ... that it linked to a dynamic lib which I did not intend to – I am using a static lib and this made me conclude that I had a path issue so it eventually linked to a wrong lib which eventually caused the issue. After fixing my paths the issue is resolved and it works perfectly fine even with the latest 1_77 Sorry for the confusion and thanks for your help ! Now it would be nice to focus on the thread priority which might possibly be wrong as well. Best Alex Hej Juan, thanks for the quick response ! >>Are you sure detach() and join() do not help? Yes, I checked it several times. I put either tracerouteThread.join() or tracerouteThread.dispatch() right after launching it but it resultet in the same effect. Please find the OSX-report here – thread 4 is the one that crashes (just in case: My app uses Qt): Thread 0:: Dispatch queue: com.apple.main-thread 0 libsystem_kernel.dylib 0x00007fff69658dfa mach_msg_trap + 10 1 libsystem_kernel.dylib 0x00007fff69659170 mach_msg + 60 2 com.apple.CoreServices.CarbonCore 0x00007fff30844d1d _scsclient_CacheableGetDataWithStringKey + 207 3 com.apple.CoreServices.CarbonCore 0x00007fff30844c44 RemoteCacheable::GetWithStringKey(unsigned int, char const*, unsigned long*, unsigned int*) + 70 4 com.apple.CoreServices.CarbonCore 0x00007fff30844ae6 _CSGetNamedData + 98 5 com.apple.AppKit 0x00007fff2c6b5e19 _NSPersistentUIGetShmem + 115 6 com.apple.AppKit 0x00007fff2c6b59c0 _NSPersistentUIEstablishTalagentCommunication + 113 7 com.apple.AppKit 0x00007fff2c6b4d91 -[NSApplication finishLaunching] + 66 8 com.apple.AppKit 0x00007fff2c6b4a20 -[NSApplication run] + 244 9 libqcocoa.dylib 0x000000010fcf2f18 QCocoaEventDispatcher::processEvents(QFlags<QEventLoop::ProcessEventsFlag>) + 2120 (qcocoaeventdispatcher.mm:431) 10 org.qt-project.QtCore 0x000000010f637d36 QEventLoop::exec(QFlags<QEventLoop::ProcessEventsFlag>) + 470 (qeventloop.cpp:232) 11 org.qt-project.QtCore 0x000000010f62f092 QCoreApplication::exec() + 130 12 com.yourcompany.soundjack 0x000000010dfb6b41 main + 2817 13 libdyld.dylib 0x00007fff69517cc9 start + 1 Thread 1: 0 libsystem_pthread.dylib 0x00007fff69717b68 start_wqthread + 0 Thread 2: 0 libsystem_pthread.dylib 0x00007fff69717b68 start_wqthread + 0 Thread 3: 0 libsystem_pthread.dylib 0x00007fff69717b68 start_wqthread + 0 Thread 4 Crashed: 0 libboost_thread.dylib 0x000000010e4a4570 tls_destructor + 96 1 libboost_thread.dylib 0x000000010e4a5784 thread_proxy + 148 2 libsystem_pthread.dylib 0x00007fff6971c109 _pthread_start + 148 3 libsystem_pthread.dylib 0x00007fff69717b8b thread_start + 15 Thread 4 crashed with X86 Thread State (64-bit): rax: 0x00000010018da37d rbx: 0x00000010018da37d rcx: 0x0000000000000008 rdx: 0x00007fef2dd18320 rdi: 0x00007fef2dd18180 rsi: 0x00000000000120a8 rbp: 0x000070000346ff70 rsp: 0x000070000346ff30 r8: 0x00000000000130a8 r9: 0x0000000000000000 r10: 0x00007fff8fc0dbf8 r11: 0x00007fff8fc0dbf0 r12: 0x00007fef2dd18180 r13: 0x0000000000000000 r14: 0x00007fef2dd18328 r15: 0x0000000000000000 rip: 0x000000010e4a4570 rfl: 0x0000000000010206 cr2: 0x00000010018da385 -- http://www.carot.de Email : Alexander(a)Carot.de Tel.: +49 (0)177 5719797 > Gesendet: Donnerstag, 07. Oktober 2021 um 21:06 Uhr > Von: "Juan Ramírez" <jramirez.uy(a)gmail.com> > An: boost-users(a)lists.boost.org > Cc: "Alexander Carôt" <alexander_carot(a)gmx.net> > Betreff: Re: [Boost-users] Thread crash in destructor > > The behavior changed in BOOST_THREAD_VERSION 3 (which is default in > boost 1.77). Are you sure detach() and join() do not help? > > Can you provide a call stack? > > On 10/7/21 06:09, Alexander Carôt via Boost-users wrote: > > Hi all, > > > > I had been using an old boost system and boost thread lib for a couple of years without upgrading it since approx. 2014. With this old lib the following (simplified) code works fine on OSX: > > > > #include "test.h" > > > > test::test() { > > boost::thread::attributes attrs; > > > > /// START TRACEROUTE THREAD > > #if defined(BOOST_THREAD_PLATFORM_WIN32) > > res = > > SetThreadPriority(attrs.native_handle(), THREAD_PRIORITY_NORMAL); > > #elif defined(BOOST_THREAD_PLATFORM_PTHREAD) > > pthread_attr_setschedpolicy(attrs.native_handle(), SCHED_FIFO); > > #endif > > tracerouteThread = boost::thread( > > attrs, boost::bind(&test::performTraceroute, this)); > > } > > > > void test::performTraceroute() { > > cout << "START TRACEROUTE" << endl; > > > > string exec = ""; > > cout << "DO FURTHER STUFF BUT THIS IS JUST A DEMO ..." << endl; > > cout << "THREAD OVER" << endl; > > } > > > > Using the current boost 1_77 still does execute the thread but then it crashes within the destructor. Adding > > > > tracerouteThread.join() or .detach() after launching the thread does not help either. > > > > Can anyone help how to resolve this issue ? > > > > Thanks a lot in advance, > > best > > > > Alex > > > > -- > > http://www.carot.de > > Email : Alexander(a)Carot.de > > Tel.: +49 (0)177 5719797 > > > > _______________________________________________ > > Boost-users mailing list > > Boost-users(a)lists.boost.org > > https://lists.boost.org/mailman/listinfo.cgi/boost-users > > >

1 0

Build issue on OSX boost 1.77 json lib is not build.
by Jiri Hoogland 07 Oct '21

07 Oct '21

Hi, I have a mysterious issue where on OSX I try to build boost 1.77 with among others the json component but it is not built. On linux there is no issue. The same script builds everything fine. Has anybody come across this by any chance ? I could not find anything from a search on the net. Thanks, Jiri Below are more details. Using the commands listed below in the OSX case, the json lib is not built, and on Linux it is. Switching on more verbosity and using the -a -q -d 5 shows that json is in the output, but not clear why it is not built. This also seems to be the case for 1.76, we tried this on multiple machines. export ZLIB_DIR=/tmp/zlib-1.2.11 export BZIP2_DIR=/tmp/bzip2-1.0.6 ./b2 link=static variant=release cxxflags="-fPIC" cflags="-fPIC" --build-type=minimal -j8 --build-dir=/tmp/boost/build stage --stagedir=/tmp/boost/stage -sBZIP2_SOURCE=${BZIP2_DIR} -sZLIB_SOURCE= ${ZLIB_DIR} -sNO_ZSTD=1 --with-filesystem --with-iostreams --with-regex --with-system --with-json ls -l /tmp/boost/stage/lib # osx 10.15.7 clang 12 cmake 3.21.3 $ uname -a Darwin vola-imac-jiri.local 19.6.0 Darwin Kernel Version 19.6.0: Tue Aug 24 20:28:00 PDT 2021; root:xnu-6153.141.40~1/RELEASE_X86_64 x86_64 $ c++ --version Apple clang version 12.0.0 (clang-1200.0.32.29) Target: x86_64-apple-darwin19.6.0 Thread model: posix InstalledDir: /Library/Developer/CommandLineTools/usr/bin $ cmake --version cmake version 3.21.3 $ ll /tmp/boost/stage/lib total 2168 drwxr-xr-x 12 jiri wheel 384 Oct 4 09:10 cmake/ -rw-r--r-- 1 jiri wheel 42688 Oct 4 09:10 libboost_atomic.a -rw-r--r-- 1 jiri wheel 85696 Oct 4 09:10 libboost_bzip2.a -rw-r--r-- 1 jiri wheel 207752 Oct 4 09:10 libboost_filesystem.a -rw-r--r-- 1 jiri wheel 156984 Oct 4 09:10 libboost_iostreams.a -rw-r--r-- 1 jiri wheel 486752 Oct 4 09:10 libboost_regex.a -rw-r--r-- 1 jiri wheel 880 Oct 4 09:10 libboost_system.a -rw-r--r-- 1 jiri wheel 116864 Oct 4 09:10 libboost_zlib.a $ ssh docs # lin gcc10 $ uname -a Linux 09964f3495a3 5.10.25-linuxkit #1 SMP Tue Mar 23 09:27:39 UTC 2021 x86_64 x86_64 x86_64 GNU/Linux $ c++ --version c++ (GCC) 10.3.0 Copyright (C) 2020 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. $ cmake --version cmake version 3.20.5 $ ls -l /tmp/boost/stage/lib total 2316 drwxr-xr-x 13 vola wheel 4096 Oct 4 12:19 cmake -rw-r--r-- 1 vola wheel 14472 Oct 4 12:19 libboost_atomic.a -rw-r--r-- 1 vola wheel 171174 Oct 4 12:19 libboost_container.a -rw-r--r-- 1 vola wheel 516560 Oct 4 12:19 libboost_filesystem.a -rw-r--r-- 1 vola wheel 226418 Oct 4 12:19 libboost_iostreams.a *-rw-r--r-- 1 vola wheel 478604 Oct 4 12:19 libboost_json.a* -rw-r--r-- 1 vola wheel 797230 Oct 4 12:19 libboost_regex.a -rw-r--r-- 1 vola wheel 1412 Oct 4 12:19 libboost_system.a -rw-r--r-- 1 vola wheel 146394 Oct 4 12:19 libboost_zlib.a

1 0

[Boost.Asio] Segmentation fault when ::getaddrinfo() returns EAI_SYSTEM without system error code set (errno = 0) - is this a bug?
by Nils Frielinghaus 04 Oct '21

04 Oct '21

Hello Boost users, I experienced a segmentation fault inside Boost in the following scenario: o There was an illegal DNS entry o When resolving that illegal DNS entry on a Yocto-based embedded Linux distribution, ::getaddrinfo() returned with the error code EAI_SYSTEM but errno was set to 0 o The function translate_addrinfo_error(int) in boost/asio/detail/impl/socket_ops.ipp translates this into boost::system::error_code(0, boost::asio::error::get_system_category()) o This is then interpreted as a success by the calling function and the other output parameters of ::getaddrinfo() are accessed which resulted in the segmentation fault My question would be whether this should be considered a bug in the Boost ASIO library? One could argue that the system should not return EAI_SYSTEM with errno not being set properly. But one could also argue that returning EAI_SYSTEM indicates that something went wrong and that the output parameters are not to be accessed. Nils ------ Nils Frielinghaus

4 11

Problem understanding 'boost::intrusive::list_member_hook<>'
by John Emmas 14 Sep '21

14 Sep '21

I'm in the early stages of trying to understand boost::intrusive::list' so forgive me if I'm just doing something stupid!! Using VS2019 I wrote a small demo app which uses 'boost::intrusive::list_base_hook<>':- #include <boost/intrusive/list.hpp> #include <string> #include <iostream> using namespace std; class animal : public boost::intrusive::list_base_hook<> { public: animal (string n, int l) : name{move(n)}, legs{l} {} string name; int legs; }; using animal_list = boost::intrusive::list<animal>; template<typename T> void print_list (const boost::intrusive::list<T>& elem) { for (const T& a : elem) cout << "A " << a.name << " has " << a.legs << " legs" << '\n'; } int main() { animal a1{"dog", 4}; animal a2{"spider", 6}; animal a3{"budgie", 2}; animal_list animals; animals.push_back(a1); animals.push_back(a2); animals.push_back(a3); print_list (animals); return 0; } Everything runs fine and I see the expected output but I've noticed that some other developers use something called a 'list_member_hook' rather than my 'list_base_hook' - so class animal would look like this:- class animal { public: animal (string n, int l) : name{move(n)}, legs{l} {} string name; int legs; boost::intrusive::list_member_hook<> _animal_hook; }; and then later on in the program there'd be a couple of lines looking something like this:- typedef boost::intrusive::member_hook<animal, boost::intrusive::list_member_hook<>, &animal::_animal_hook> AnimalHookOption; typedef boost::intrusive::list<animal, AnimalHookOption> Animals; But when I try that here, VS2019 throws up this compiler error:- error C2039: 'default_list_hook': is not a member of 'animal' Can anyone explain what I'm doing wrong?? John

4 18

Support end date of boost version ,1.76.0
by rajesh daggumati 08 Sep '21

08 Sep '21

Hi Team, When is supporting end date for boost 1.76.0. Please let me know asap. Regards, Rajesh D

2 1

Linking Problem
by Stephen P. Molnar 04 Sep '21

04 Sep '21

I have successful compiled an application on my Debian-10 platform using boost-1-67. It is an essential part of my computations chemistry pipe line. Now Debian 11 has been released. I have installed it in a VM in order to make sure that the app is compilable in Debian 11 with boost-1-74 (the Debian12 version of boost), before upgrading my Linux platform. I installed boost in /user/ilocal. The makefile is: BASE=/usr/local BOOST_VERSION=1_74_0 BOOST_INCLUDE = $(BASE)/include C_PLATFORM= -pthread GPP=g++ C_OPTIONS= -O3 -DNDEBUG BOOST_LIB_VERSION= include ../../makefile_common The test compile went well until the link stage. The process stopped with: g++ -pthread -ansi -Wno-long-long -O3 -DNDEBUG -I /usr/local/include -L/usr/local/lib -L. -o gwovina main.o cache.o coords.o current_weights.o everything.o grid.o szv_grid.o manifold.o model.o gwo.o gwo_mutate.o my_pid.o naive_non_cache.o non_cache.o parallel_gwo.o parse_pdbqt.o pdb.o quasi_newton.o quaternion.o random.o ssd.o terms.o weighted_terms.o rotamer.o -l boost_system -l boost_thread -l boost_program_options /usr/bin/ld: cannot find -lboost_system /usr/bin/ld: cannot find -lboost_thread /usr/bin/ld: cannot find -lboost_program_options collect2: error: ld returned 1 exit status make: *** [../../makefile_common:34: gwovina] Error 1 Even my limited knowledge (my original programming experience was with FORTRAN II in the early 1960's), I can see the the problem. Unfortunately, I don't know the solution. Assistance will be greatly appreciated. Thanks om advance. -- Stephen P. Molnar, Ph.D. www.molecular-modeling.net 614.312.7528 (c) Skype: smolnar1

2 1

Comment on divide_unsigned_helper
by Neill Clift 03 Sep '21

03 Sep '21

Hi, I was trying to understand the algorithm inside the main divide routine. I struggled with this bit of code: if ((prem[r_order] <= py[y_order]) && (r_order > 0)) { double_limb_type a = (static_cast<double_limb_type>(prem[r_order]) << CppInt1::limb_bits) | prem[r_order - 1]; double_limb_type b = py[y_order]; double_limb_type v = a / b; if (v <= CppInt1::max_limb_value) { guess = static_cast<limb_type>(v); --r_order; } } I was trying to understand why we essential set guess to 1 in the case that the division is too big. I finally realized that the only way we can have v > CppInt1::max_limb_value is if we have prem[r_order] == py[y_order]. Some simple inequalities with floor can convince you of this. So in the case of equality we do an expensive divide and test when we could move the test outside like this: limb_type guess = 1; if (r_order == 0) { <-------- do this first to simplify guess = prem[0] / py[y_order]; } else if (prem[r_order] == py[y_order]) { <----------- handle it here to avoid the divide guess = 1; ... This leads to another question. Why is this line: guess = prem[0] / py[y_order]; not this line: guess = prem[0] / py[0]; Logically for division it must be the second line. In fact this loop has the invariant that r_order >= y_order. That makes sense for division. So the divisor must be py[0]. That suggests a divisor with one element and we handle that case via a special divide. So that code is not needed I believe. We end up with this: BOOST_ASSERT(r_order >= y_order && y_order >= 1); if (prem[r_order] == py[y_order]) { guess = 1; } else if (prem[r_order] < py[y_order]) { double_limb_type a = (static_cast<double_limb_type>(prem[r_order]) << CppInt1::limb_bits) | prem[r_order - 1]; double_limb_type b = py[y_order]; double_limb_type v = a / b; BOOST_ASSERT(v <= CppInt1::max_limb_value); guess = static_cast<limb_type>(v); --r_order; } else { double_limb_type a = (static_cast<double_limb_type>(prem[r_order]) << CppInt1::limb_bits) | prem[r_order - 1]; double_limb_type b = (y_order > 0) ? (static_cast<double_limb_type>(py[y_order]) << CppInt1::limb_bits) | py[y_order - 1] : (static_cast<double_limb_type>(py[y_order]) << CppInt1::limb_bits); BOOST_ASSERT(b); double_limb_type v = a / b; guess = static_cast<limb_type>(v); } BOOST_ASSERT(guess); // If the guess ever gets to zero we go on forever.... Some simple random divide tests seem to show a little over 2% improvement from doing this. Of course this could just be code layout changes. Just thinking that the 64 bit values are random would make me think this case is rare but I guess we subtract out multiples of the divisor and cause this to happen more for reasons that are hard to understand (for me anyway). I see the path hit a lot. Neill.

1 0