May 2009 - Boost-users - lists.boost.org

boost.python error while ./bjam install
by Aamir Ahmed Khan 08 May '09

08 May '09

I am compiling boost.python libraries by following the procedure given at http://www.boost.org/doc/libs/1_39_0/more/getting_started/unix-variants.htm… I first configure the build and installation using bootstrap.sh script $ ./bootstrap.sh --with-python=/usr/bin/python2.5 --with-libraries=python --prefix=~/work/boost/setup/boost_1_39_0/mybuild Here are the contents of project-config.jam fie as generated by the script. # Boost.Build Configuration > # Automatically generated by bootstrap.sh > > import option ; > import feature ; > > # Compiler configuration. This definition will be used unless > # you already have defined some toolsets in your user-config.jam > # file. > if ! gcc in [ feature.values <toolset> ] > { > using gcc ; > } > > project : default-build <toolset>gcc ; > > # Python configuration > using python : 2.5 : /usr ; > > # List of --with-<library> and --without-<library> > # options. If left empty, all libraries will be built. > # Options specified on the command line completely > # override this variable. > libraries = --with-python ; > > # These settings are equivivalent to corresponding command-line > # options. > option.set prefix : ~/work/boost/setup/boost_1_39_0/mybuild ; > option.set exec-prefix : ~/work/boost/setup/boost_1_39_0/mybuild ; > option.set libdir : /lib ; > option.set includedir : ~/work/boost/setup/boost_1_39_0/mybuild/include > ; > The output of bootstrap.sh script instructs me to run ./bjam to build while instructions at the above link say to run ./bjam install Anyways, I tried both the options but every time I get the same error > Class top-level-target already defined > Aborted > I searched the forums, but this error seems to never be discussed. I am stuck, please help! Best regards, Aamir Ahmed Khan

1 0

Statically linking libstdc++.so.6 and boost
by Michael Dehmlow 08 May '09

08 May '09

I have a rather interesting situation and I'm wondering if any one has some ideas. I am building a linux shared library which uses boost and xerces, I have an application which I compile and run fine using gcc 4.1 which uses stdlibc++.so.6. I also have a requirement to create an interface for Labview 8.6 which is built using gcc 3.3 which uses stdlibc++.so.5. After some investigation I found that Labview would crash on any std::string concatenation because it was dynamically using the old stdlib(5). In a simplified test tool I statically linked the stdc++ lib using this really great page: http://www.trilithium.com/johan/2005/06/static-libstdc/ Everything worked great until I started adding my cots (boost and xerces) into the mix. For xerces I simply modified the xerces make file to statically link stdlib and (using gdb was able determine that) I get past all my xerces calls. However when I make call to boost lexical cast it still tries to use the dynamic std library 5: #0 0x00cab1a7 in std::string::data () from /usr/lib/libstdc++.so.5 #1 0x0362b479 in boost::detail::lexical_cast<int, std::string, false, char> (arg=@0x0, buf=0x0, src_len=0) at /usr/include/boost/boost_1_38_0/boost/lexical_cast.hpp:1149 #2 0x0362b500 in boost::detail::lexical_cast<int, std::string, false, char> (arg=Cannot access memory at address 0x8 ) at /usr/include/boost/boost_1_38_0/boost/lexical_cast.hpp:1159 Above is the stack trace immediately preceding my segmentation fault. One: I thought that lexical cast was built into hpp files and was compiled at my application compile time is that correct? Two: does any one know how to make the symbols from a .a (static library) file look in the encapsulating so's symbol table before the applications symbol table? Or can anyone point me at some information on symbol lookup precedence. Three: I'm considering including all of the generated object files from the included boost .a files in my make file rather than including the staticboost libraries themselves. Seem like a good idea? Any pitfalls? Four: any other Ideas? Below is the meat of my make file: g++ -g -static-libgcc -nodefaultlibs -fPIC -shared -fno-gnu-linker -Wl,-soname,libEXPO.so -o libEXPO.so ${APIOBJECTS} ${DETAILOBJECTS} ${OBJECTS} -L. -L../../../utl/CommonObjects/XML/ -L../../../utl/CommonObjects/Exceptions ${COTS_EXPO_BOOST_LIBDIR_38}/libboost_thread-gcc41-mt-1_38.a ${COTS_EXPO_BOOST_LIBDIR_38}/libboost_filesystem-gcc41-mt-1_38.a ${COTS_EXPO_BOOST_LIBDIR_38}/libboost_system-gcc41-mt-1_38.a ${COTS_EXPO_BOOST_LIBDIR_38}/libboost_regex-gcc41-mt-1_38.a ${COTS_EXPO_BOOST_LIBDIR_38}/libboost_date_time-gcc41-mt-1_38.a ./libxerces-c.so.26 -lXML -lExceptions -lc /lib/libpthread.so. -- View this message in context: http://www.nabble.com/Statically-linking-libstdc%2B%2B.so.6-and-boost-tp234… Sent from the Boost - Users mailing list archive at Nabble.com.

1 0

[test] Crash when running my unit tests
by Christian Henning 08 May '09

08 May '09

Hi there, when I try to run my tests the unit testing framework seems to crash for no apparent reason. The point it's crashing is in unit_test_suite.ipp[263]. It seems to crash before entering the first test. The c++ runtime at this points tries to delete the memory of a string which triggers the assertion. _ASSERTE(_BLOCK_TYPE_IS_VALID(pHead->nBlockUse)); Here is my scenario: Everything used to run just fine before I added a new file containing more tests. Meaning all of my tests are scattered around a couple of files ( 20 files ) which sum up to about 120 tests. I don't think that this usage is anything but exceptional. When I exclude the new test file the tests are running fine. When including the new test file it crashes. I'm running the boost trunk on a windows xp using MSVC 2005. Anyone any idea what I can do about this? My code is available here: http://gil-contributions.googlecode.com/svn/trunk/ The tests are here: http://gil-contributions.googlecode.com/svn/trunk/gil_2/libs/gil/io_new/tes… Any help is more than welcome. Thanks ahead, Christian

4 11

Unique parameter values for interfaces
by Ryan McConnehey 08 May '09

08 May '09

I'm currently trying to fortify my class api by "make interfaces easy to use correctly and hard to use incorrectly". I have a class as follows. class api { api(ConfigName const& configName_, PathName const& pathName_) : m_ConfigName(configName_.value) , m_PathName(pathName_.value) {} std::string m_ConfigName; std::string m_PathName; }; Both ConfigName and PathName are typedef to a template structure as follows. template<typename S> struct ParameterValue { ParameterValue(S value_) : value(value_) {} S value }; typedef ParameterValue<std::string> ConfigName; typedef ParameterValue<std::string> PathName; I realized later that this method doesn't prevent the swapping of a ConfigName with a PathName parameter. Since both parameters are typedef to the same "type" of ParameterValue they're interchangeable. Is there a boost class, library or C++ idiom that would provide what I'm looking for? Since I'm going to be using this concept for other classes I build, I'd like something that would easily create parameter classes to use but not be interchangeable with other classes of the same type. I know that I could do this with a macro but I'm leaving that for a last resort. Ryan

5 9

boost mpi + pointer to polymorphic class
by Michael Levenhagen 08 May '09

08 May '09

Hello, I'm having a problem using boost::mpi::communicator::isend(). The problem is that it appears isend can't serialize via a pointer to the base of a polymorphic class. I'm using boost::serialization extensively to archive and I believe serializing via a pointer to a base class works when archiving. This is what I'm trying: class Event { public: virtual ~Event(); }; class Der1Event : Event { }; class Der2Event: Der1Event { }; Der2Event* d2 = new Der2Event; Der1Event* d1 = static_cast<Der1Event*>(d2) boost::mpi::communicator world; world.isend( 0, 0, d1 ); My program faults when calling isend(). I know my example is missing details such as serialization code but in principle should this work? BTW the following works. world.isend( 0, 0, d2 ) thanks Mike

1 0

Steps to build boost library for 64 bit platform.
by Jaiswal, Usha [OS-IE] 08 May '09

08 May '09

Hi, Can any one tell me the steps to build boost library for 64 bit platform, please reply ASAP. Thanks a lot. -Usha.

2 1

How boost::tupe avoid compilor error of get member function?
by simon Ye 08 May '09

08 May '09

Hi, all. boost::tuple provide a member template function get to gain a element of tuple, As following: #include "boost\tuple\tuple.hpp" using namespace boost; template<typename T> class TA{public: template<typename T1, typename T2> void testfun(T1 e1, T2 e2) { tuple<T1, T2> t1(e1, e2); //T1 _e1 = t1.template get<0>(); //works *T1 _e1 = t1.get<0>(); //works for boost::tuple* /* However, according c++98, t1.get<0>() seems to should be encounter a compilor error. *I want to known how boost::tuple avoid this compile error.* */ } }; int main() { int i=10; float f=10.5f; TA<char> obj; obj.testfun(i, f); return 0; } The above sample pass the compilation of gcc3.4.2 and gcc 4.2.4. *According to boost::tuple, I implement a new tuple by myself. as following: * (In fact, the following code is mostly extracted from boost::tuple. ) namespace testNS{ template<int N, class T> struct element { private: typedef typename T::tail_type Next; public: typedef typename element<N-1, Next>::type type; }; template<class T> struct element<0,T> { typedef typename T::head_type type; }; struct null_type{}; template<typename HT, typename TT> struct cons { typedef HT head_type; typedef TT tail_type; template <int N> typename element<N, cons<HT, TT> >::type *get()* { typedef typename element<N, cons<HT, TT> >::type RetType; return RetType(); } }; template<typename HT> struct cons<HT, null_type> { typedef HT head_type; typedef null_type tail_type; typedef cons<HT, null_type> self_type; //get(BOOST_EXPLICIT_TEMPLATE_NON_TYPE(int, N)) template <int N> typename element<N, self_type>::type *get()* { typedef typename element<N, cons<HT,null_type> >::type RetType; return RetType(); //return boost::tuples::get<N>(*this); } }; template <class T0, class T1, class T2> struct map_tuple_to_cons { typedef cons<T0, typename map_tuple_to_cons<T1, T2, null_type>::type > type; }; // The empty tuple is a null_type template <> struct map_tuple_to_cons<null_type, null_type, null_type> { typedef null_type type; }; template<typename T0, typename T1, typename T2=null_type> class Tuple; template<typename T0, typename T1, typename T2> class Tuple: public map_tuple_to_cons<T0, T1, T2>::type { public: typedef T0 E0T; typedef T1 E1T; typedef T2 E2T; Tuple(const E0T& e0, const E1T& e1) { } }; }//endof testNS using namespace testNS; template<typename T> class TA{public: template<typename T1, typename T2> void testfun(T1 e1, T2 e2) { Tuple<T1, T2> t1(e1, e2); //T1 _e1 = t1.template get<0>(); //works * T1 _e1 = t1.get<0>(); //encounter compilor error /* the gcc3.4.2 error message: syntax error before `;' token the gcc4.2.4 error message: error: expected primary-expression before ')' token error: invalid operands of types '<unresolved overloaded function type>' and 'int' to binary 'operator<' */* } }; int main() { int i=10; float f=10.5f; TA<char> obj; obj.testfun(i, f); return 0; } By the way, how can i get contact with boost::tuple author.? -- miss you!

2 1

Fwd: banded symmetricmatrix
by Hicham Mouline 08 May '09

08 May '09

Sending this to boost users as well as ublas's list seems a bit dry... Rds, --------------- Hello, Is there a template to store a matrix that is both band diagonal and symmetric? Is there a compact storage for it that is even more optimal than just banded's and symmetric's ? Regards,

1 0

[Date time] warning C4244
by Seweryn Habdank-Wojewódzki 08 May '09

08 May '09

Dears, I have a lot of warnings: boost/date_time/microsec_time_clock.hpp(103) : warning C4244: 'argument' : conversion from 'int' to 'unsigned short', possible loss of data Best regards, -- |\/\/| Seweryn Habdank-Wojewódzki \/\/ Professionalism in programming - www.accu.org ---------------------------------------------------- Zobacz extremalne skoki (base jumping) i jazdę na krechę - kliknij: http://klik.wp.pl/?adr=http%3A%2F%2Fcorto.www.wp.pl%2Fas%2Fe4.html&sid=720

1 0

Re: [Boost-users] CMake Problems with boost 1.39
by Tan, Tom (Shanghai) 08 May '09

08 May '09

>> - The generated boost.test libs have an additional "-s" to the file >> names, ex. libboost_unit_test_framework-vc90-mt-1_39-s.lib, instead of >> libboost_unit_test_framework-vc90-mt-1_39.lib. This will cause a linking >> error while using boost.test. this does noto happen to the bjamed >> results. >> >.-s means that the library links to the static C++ runtime. The issue is, -s is appended to the end of the file name to all .lib files, even for those statically linked files that already have 's' in the middle. Ex. libboost_unit_test_framework-vc90-mt-s-1_39-s.lib, rather than libboost_unit_test_framework-vc90-mt-s-1_39.lib

1 0