- Boost-users - lists.boost.org

[thread] Can Boost.Thread use Boost.Atomic without falling on a compatibility issue?
by Vicente J. Botet Escriba 12 Jan '13

12 Jan '13

Hi, in order to fix this issue, https://svn.boost.org/trac/boost/ticket/5752 : "boost::call_once() is unreliable on some platforms", I would like to use Boost.Atomic as suggested on the ticket. As Boost.Atomic is not a header-only library the user would need to link with boost_atomic. Is this an acceptable change? I'm requesting this as recently there were some people reactive to some changes in Boost.Thread. Would this be considered an compatibility issue? If yes should the fix for the call_once issue be included if the user request it using conditional compilation or even go on a separated version of Boost.Thread? Thanks, Vicente

1 0

Newbie Question
by Walker.Kyle 12 Jan '13

12 Jan '13

It appears the following code has a warning detected by GCC4.7.2: /boost/include/boost-1_52/boost/property_tree/detail/ptree_utils.hpp: In instantiation of 'std::string boost::property_tree::detail::narrow(const Ch*) [with Ch = wchar_t; std::string = std::basic_string<char, std::char_traits<char>, std::allocator<char> >]': /boost/include/boost-1_52/boost/property_tree/string_path.hpp:64:36: required from here /boost/include/boost-1_52/boost/property_tree/detail/ptree_utils.hpp:76:13: warning: comparison between signed and unsigned integer expressions [-Wsign-compare] boost-1_52\boost\property_tree\detail\ptree_utils.hpp (Starting line 70): template<class Ch> std::string narrow(const Ch *text) { std::string result; while (*text) { if (*text < 0 || *text > (std::numeric_limits<char>::max)()) Shouldn't it be: template<class Ch> std::string narrow(const Ch *text) { std::string result; while (*text) { if (*text < 0 || *text > (std::numeric_limits<Ch>::max)()) Code to reproduce this warning (just the include): #include <boost/property_tree/ptree.hpp>

2 1

Boost and -fvisibility=hidden
by Dániel Varga 11 Jan '13

11 Jan '13

Hi, I would like to add symbol visibility support to my library. Using -fvisibility=hidden option I compared the changed symbols, boost::bad_function_call typeinfo was visible previously but with hidden default visibility it is hidden. It compiles just fine, but according to gcc wiki (http://gcc.gnu.org/wiki/Visibility) exceptions must be visible as runtime error might happen because of missing typeinfo. I found bug report (https://svn.boost.org/trac/boost/ticket/2309) that exception class not exported in Boost.Thread, the problem was fixed by adding pragmas. In boost/function/function_base.hpp pragmas or other ways to change visibility isn't used, so the boost::bad_function_call typeinfo became hidden using -fvisibility=hidden. In the former boost bug report (https://svn.boost.org/trac/boost/ticket/2309) runtime error does not even mentioned, I couldn't reproduce the runtime error in my small experiments. In what situations can hidden symbol cause problems? Is it normal that boost::bad_function_call typeinfo is hidden or it is a boost bug that can cause problems and should be fixed? A made a small test library and application. The library has hidden boost::bad_function_call typeinfo, but the application seems catching the exception properly. Can this cause an error, why is it working? Makefile: CXX=g++ test_exception: test_exception.cpp libmydso.so $(CXX) -L. -lmydso $< -o $@ libmydso.so: mydso.cpp $(CXX) -fvisibility=hidden -fPIC -shared -Wl,-soname,$@ $< -o $@ mydso.h: void __attribute__ ((visibility ("default"))) test_exception(); mydso.cpp: #include <boost/function.hpp> #include "mydso.h" void test_exception() { boost::function<int (int, int)> f; f(5, 4); } test_exception.cpp: #include <iostream> #include <stdexcept> #include "mydso.h" int main() { try { test_exception(); } catch (std::runtime_error) { std::cerr << "Got std::runtime_error" << std::endl; } catch (...) { std::cerr << "Got other exception" << std::endl; } } Thanks, Dani

1 0

Re: [Boost-users] mixed 32 bit and 54 bit
by Dave Dyer 11 Jan '13

11 Jan '13

> > >You need to define an offset_ptr with 64 bit OffsetType > >typedef offset_ptr<T, std::ptrdiff_t, unsigned long long> my_offset_ptr; > >and instantiate all interprocess managed resources like >(basic_managed_shared_memory) like this: > > >typedef basic_managed_shared_memory > <char > ,rbtree_best_fit<mutex_family, my_offset_ptr> > ,iset_index> >managed_shared_memory; > >Again, this is not supported and it's still experimental. Thanks for the guidance. It may work for my pirposes, as I dont intend to use any boost facilities except some basic block allocations of memory. One suggestion; make a 32-vs-64 bit test part of the setup code, so when a known incompatible producer/consumer relationship exists, it can gracefully throw an error instead of failing mysteriously.

1 0

boost::range vs && (rvalue ref)
by Neal Becker 11 Jan '13

11 Jan '13

Passing range by ref (&) is fine, but using (&&) results in and error. Any ideas? #include <boost/range.hpp> #include <vector> template<typename out_t> void F (out_t && o) { typedef typename boost::range_value<out_t>::type T; } void G() { std::vector<int> v; F (v); } usr/local/src/boost_1_52_0/boost/mpl/eval_if.hpp:60:31: error: no type named ‘type’ in ‘boost::mpl::eval_if_c<false, boost::range_const_iterator<std::vector<int>&>, boost::range_mutable_iterator<std::vector<int>&> >::f_ {aka struct boost::range_mutable_iterator<std::vector<int>&>}’ bug.cc: In instantiation of ‘void F(out_t&&) [with out_t = std::vector<int>&]’: bug.cc:12:7: required from here bug.cc:6:52: error: no type named ‘type’ in ‘struct boost::range_value<std::vector<int>&>’

3 4

Re: [Boost-users] mixed 32bit and 64bit shared memory
by Dave Dyer 11 Jan '13

11 Jan '13

I see notes about boost supporting "customizing offset_ptr" to support shared memory where both 64 bit and 32 bit processes are sharing the memory, but I don't see a specific example of how to do that. It seems that "out of box" it doesn't work.

2 1

Re: [Boost-users] mixed 32bit and 64bit shared memory
by Dave Dyer 11 Jan '13

11 Jan '13

I see notes about boost supporting "customizing offset_ptr" to support shared memory where both 64 bit and 32 bit processes are sharing the memory, but I don't see a specific example of how to do that. It seems that "out of box" it doesn't work.

1 0

[1.53.0] Beta release candidates available
by Marshall Clow 10 Jan '13

10 Jan '13

Release candidate files for 1.53.0 beta 1 are available at 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: d5a45d0e08a14d5fb0965fe675d76ac6 *boost_1_53_0_beta1_rc1.tar.gz 89a6084458cd9f0e136c4b5292265461 *boost_1_53_0_beta1_rc1.tar.bz2 2a8897e04c59ccaf39189568f6966d60 *boost_1_53_0_beta1_rc1.zip 4cdf5a71d46e1115ca0c413f5b139a54 *boost_1_53_0_beta1_rc1.7z Thanks, --The release managers

4 5

build without python
by Philipp Kraus 10 Jan '13

10 Jan '13

Hello, I try to build Boost with MPI on a system and I get errors, because Boost try to build python also. I run b2 and I build with the --with option the following libraries Performing configuration checks - gcc visibility : yes - long double support : yes - has_icu builds : no Component configuration: - chrono : not building - context : not building - date_time : building - exception : building - filesystem : building - graph : not building - graph_parallel : not building - iostreams : building - locale : not building - math : building - mpi : building - program_options : building - python : not building - random : building - regex : building - serialization : building - signals : not building - system : building - test : not building - thread : building - timer : not building - wave : not building The b2 command is: --with-exception --with-math --with-system --with-iostreams --layout=system link=shared runtime-link=shared threading=multi variant=release install --prefix=../build_release/boost/1.52.0 --with-regex --with-random --with-thread --with-date_time --with-program_options --with-filesystem --with-mpi --with-serialization The build process breaks with the error eg: ...failed gcc.compile.c++ bin.v2/libs/python/build/gcc-4.6/release/threading-multi/long.o... gcc.compile.c++ bin.v2/libs/python/build/gcc-4.6/release/threading-multi/dict.o In file included from ./boost/python/detail/prefix.hpp:13:0, from ./boost/python/dict.hpp:8, from libs/python/src/dict.cpp:4: ./boost/python/detail/wrap_python.hpp:75:24: fatal error: patchlevel.h: No such file or directory compilation terminated. If I mix --with and --without option on the b2 command I get the error: error: both --with-<library> and --without-<library> specified How can I disable any Python builds? Thanks Phil

2 3

build without python
by Philipp Kraus 10 Jan '13

10 Jan '13

Hello, I try to build Boost with MPI on a system and I get errors, because Boost try to build python also. I run b2 and I build with the --with option the following libraries Performing configuration checks - gcc visibility : yes - long double support : yes - has_icu builds : no Component configuration: - chrono : not building - context : not building - date_time : building - exception : building - filesystem : building - graph : not building - graph_parallel : not building - iostreams : building - locale : not building - math : building - mpi : building - program_options : building - python : not building - random : building - regex : building - serialization : building - signals : not building - system : building - test : not building - thread : building - timer : not building - wave : not building The b2 command is: --with-exception --with-math --with-system --with-iostreams --layout=system link=shared runtime-link=shared threading=multi variant=release install --prefix=../build_release/boost/1.52.0 --with-regex --with-random --with-thread --with-date_time --with-program_options --with-filesystem --with-mpi --with-serialization The build process breaks with the error eg: ...failed gcc.compile.c++ bin.v2/libs/python/build/gcc-4.6/release/threading-multi/long.o... gcc.compile.c++ bin.v2/libs/python/build/gcc-4.6/release/threading-multi/dict.o In file included from ./boost/python/detail/prefix.hpp:13:0, from ./boost/python/dict.hpp:8, from libs/python/src/dict.cpp:4: ./boost/python/detail/wrap_python.hpp:75:24: fatal error: patchlevel.h: No such file or directory compilation terminated. If I mix --with and --without option on the b2 command I get the error: error: both --with-<library> and --without-<library> specified How can I disable any Python builds? Thanks Phil

1 0