- Boost-users - lists.boost.org

libboost_date_time: undefined reference to __sync_fetch_and_add_4
by Carsten Raas 07 Apr '09

07 Apr '09

Hello! I built a (complete) Boost 1.38 library with the Intel compilers 10.1.021, 11.0.073, and 11.0.081 for an EM64T (x86_64) Linux architecture. But I end up with the same problem for all builds: the shared version of libboost_date_time cannot be linked to this simple demo program: # cat conftest.cpp ----------------------------------------------------------------- #include <boost/date_time/date_generators.hpp> int main () { const char *ptr = boost::date_time::nth_as_str(0); return 0; } ----------------------------------------------------------------- # icpc -V Intel(R) C++ Intel(R) 64 Compiler Professional for applications running on Intel(R) 64, Version 11.0 Build 20090131 Package ID: l_cproc_p_11.0.081 ----------------------------------------------------------------- Static linking: # icpc -I /opt/boost/1.38/intel/include conftest.cpp \ /opt/boost/1.38/intel/lib/libboost_date_time.a => Works. ----------------------------------------------------------------- Shared linking: # icpc -I /opt/boost/1.38/intel/include conftest.cpp -L /opt/boost/1.38/intel/lib -lboost_date_time /opt/boost/1.38/intel/lib/libboost_date_time.so: undefined reference to `__sync_fetch_and_add_4' ----------------------------------------------------------------- Shared linking with gcc: # g++ -I /opt/boost/1.38/intel/include conftest.cpp \ -L /opt/boost/1.38/intel/lib -lboost_date_time /opt/boost/1.38/intel/lib/libboost_date_time.so: undefined reference to `__sync_fetch_and_add_4' collect2: ld returned 1 exit status ----------------------------------------------------------------- Shared linking with gcc to libraries compiled with gcc itself: # g++ -I /opt/boost/1.38/gcc/include conftest.cpp \ -L /opt/boost/1.38/gcc/lib -lboost_date_time => Works. ----------------------------------------------------------------- I have been googling around for a while, but all I found was related to GCC sync primitives (e.g. sync_lock_test_and_set, sync_fetch_and_add) and different architectures (armel, sparc, etc.). I also tried the workaround with #define BOOST_SP_USE_PTHREADS in boost/config/user.hpp, without success. I also checked 1.37 and today's SVN version, but they show the same problems. Any hints what I should look at next? Best regards, -Carsten

4 9

[threads] recursive shared_mutex
by joaquin＠tid.es 07 Apr '09

07 Apr '09

Hi, As the docs do not explicitly say otherwise, I guess I must assume that boost::shared_mutex is not recursive with respect to unique ownership, that is, the following behavior ensues: shared_mutex mutex; lock_guard<shared_mutex> lock1(mutex); lock_guard<shared_mutex> lock2(mutex); // DEADLOCK Is this so? In case boost::shared_mutex is indeed non-recursive, I've read the following assertion on Hinnant's paper N2406 (http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2007/n2406.html ): "A recursive shared mutex could be built upon shared_mutex if desired, but this is not proposed." Does anyone has a reference on how to construct a recursive shared_mutex from a non-recursive one? Thank you, Joaquín M López Muñoz Telefónica, Investigación y Desarrollo

2 1

Boost, GCC, TR1 question
by Anthony Foglia 07 Apr '09

07 Apr '09

We've stumbled across an incompatiblity in mixing TR1 stuff from GCC 4.3 and Boost. The smallest test case is simply #include <tr1/array> #include <boost/tr1/tuple.hpp> This fails because GCC's array declares and specializes the GCC's tuple_size and tuple_element, while boost/tr1/tuple.hpp by default uses ::boost::fusion::tuple_size and tuple_element. It appears that defining BOOST_HAS_GCC_TR1 before including the boost libraries will solve the problem. Is there a better way? More importantly, this is code in part of library used by a number of other projects. Will we have to recompile the boost libraries (we're using the current Ubuntu package), and/or any of our projects that link against this code, or should linkage not be a problem? -- Anthony Foglia Princeton Consultants (609) 987-8787 x233

2 1

Boost.spirit: how to handle file parse errors?
by Joost Kraaijeveld 07 Apr '09

07 Apr '09

Hi, I am parsing a large file and experience errors. Is there an example available somewhere that shows how to print something like "error at file, line, mismatched string" somehow? TIA -- Groeten, Joost Kraaijeveld Askesis B.V. Molukkenstraat 14 6524NB Nijmegen tel: 024-3888063 / 06-51855277 fax: 024-3608416 web: www.askesis.nl

1 0

Selecting among different MPI versions
by Alain Leblanc 07 Apr '09

07 Apr '09

Hi, I'm trying to build boost 1.38 with MPI support. The problem is that my system has more than one version of MPI installed, and the boost build uses the wrong one. I don't know how boost decides which one to use. The following are the only default defined by gcc, and none of them contains mpi.h g++ -Wp,-v -E - ignoring nonexistent directory "/usr/lib/gcc/x86_64-redhat-linux/4.1.1/../../../../x86_64-redhat-linux/include" #include "..." search starts here: #include <...> search starts here: /usr/local/include /usr/lib/gcc/x86_64-redhat-linux/4.1.1/include /usr/include End of search list. When I try building boost, I get a bunch of messages of the form ...failed gcc.compile.c++ bin.v2/libs/mpi/build/gcc-4.1.1/release/threading-multi/graph_communicator.o... gcc.compile.c++ bin.v2/libs/mpi/build/gcc-4.1.1/release/threading-multi/group.o g++: /opt/intel/mpi-rt/3.1: No such file or directory g++: unrecognized option '-rpath' In file included from ./boost/mpi/config.hpp:20, from ./boost/mpi/exception.hpp:16, from ./boost/mpi/group.hpp:18, from libs/mpi/src/group.cpp:9: /usr/local/intel/impi/3.1/include/mpi.h:35:8: error: #error The wrong version of mpi.h file was included, check include path. and the version of MPI I want to use is in /usr/local/mpich/... So is there some option to tell boost which mpi.h to include. I tried setting the CPLUS_INCLUDE_PATH variable, but it didn't change anything. I have no control over the OS itself. Thanks, a

1 0

Intrusive list/ilist and rbtree marshalling support
by Brian Wood 07 Apr '09

07 Apr '09

I'm happy to announce that version 1.8 of the C++ Middleware Writer is available on line -- http://webEbenezer.net/cgi-bin/samb.cgi<http://webebenezer.net/cgi-bin/samb.cgi>. This version adds support for the following: * Exclusive marshalling support of boost::intrusive::list and boost::intrusive::rbtree. Neither Boost Intrusive nor Boost Serialization offer serialization support of the Intrusive containers. Rather than having two types called 'list,' we refer to boost::intrusive::list as an ilist. An example of this support is here -- http://webEbenezer.net/rbtree_marshalling.html<http://webebenezer.net/rbtree_marshalling.html>. * Several new options on the user interface including: + An "Error Handling" option that indicates whether the code should be written using return codes or exceptions. The exception support is new in this release. This option affects the signatures of the generated marshalling functions. + A "Code Structure" option that indicates whether the code should be written in header-only format or separated interface and implementation files. The separated interface and implementation support is new in this release. This option also affects the signatures of the generated marshalling functions. + A choice between unsigned short or unsigned int as the type for constants written by the C++ Middleware Writer. Constants that identify messages and user defined types are written by the software. Previously the constants were always unsigned ints. * A mechanism to turn off automatic code generation for a type. Adding a // comment within a class declaration with "hand_written_marshalling_code" in the text of the comment will disable default code generation for that type. For example, class bag { // hand_written_marshalling_code ... }; As the comment indicates, you will provide the implementation of the marshalling functions for this class. This option is not available with /* */ comments. Additionally, I would like to say that I find the Intrusive library to be sound and am using it internally in the C++ Middleware Writer. Thanks to the authors: Olaf Krzikalla and Ion Gaztañaga. Brian Wood Ebenezer Enterprises www.webEbenezer.net <http://www.webebenezer.net/> "Then Samuel took a rock and set it up between Mizpah and Shen. He named it Ebenezer [Rock of Help] and said, 'Until now the L-RD has helped us.'"

1 0

Matching constant input with input streams
by Eugene Wee 06 Apr '09

06 Apr '09

Hi, A fellow user on an online forum that I frequent asked about the use of C++ input streams to replace the use of scanf() and family where the format string included constant portions, e.g., the "a=" in "a=%d". The answer was to read the input as a string and then parse it. Another user chimed in to say that he preferred sscanf() for this reason, and that C++ input streams would be more useful if such constant portions could be easily matched. I wrote a manipulator that did just that as a proof of concept, but now I wonder if Boost already provides such a manipulator, or if say, Boost.Spirit with its file_iterator is deemed as the right solution. There certainly seems to be a trade-off between these two choices: a manipulator would be a simple add-on to existing standard library functionality, but Boost.Spirit is much more powerful. Thank you, Eugene Wee

1 0

[range_ex] range_ex conflicts with non_copyable
by 041021042 06 Apr '09

06 Apr '09

Hi， The following code works well with old version of range_ex (Dec 2008), but when I update range_ex to an newer version (after Feb 2009), the code cannot compile. code: #include <boost/range/algorithm.hpp> #include <boost/ptr_container/ptr_vector.hpp> struct A : boost::noncopyable { int a; A(int _a) : a(_a) {} bool operator == (const A& r) const { return a == r.a; } }; int main() { boost::ptr_vector<A> pva; A a(0); std::find(pva.begin(), pva.end(), a); // ok boost::find(pva, a); // cause an error } compiler: msvc2008 error message: e:\zxg\boost\range_ex\range.cpp(8) : error C2248: 'boost::noncopyable_::noncopyable::noncopyable' : cannot access private member declared in class 'boost::noncopyable_::noncopyable' d:\software\boost_1_37_0\boost\noncopyable.hpp(27) : see declaration of 'boost::noncopyable_::noncopyable::noncopyable' d:\software\boost_1_37_0\boost\noncopyable.hpp(22) : see declaration of 'boost::noncopyable_::noncopyable' This diagnostic occurred in the compiler generated function 'A::A(const A &)' Thanks, Roger Zhou

5 5

Reuse of boost config and other related headers
by Seweryn Habdank-Wojewódzki 06 Apr '09

06 Apr '09

Dears, How could I reuse boost config and neighbour to prepare platform independent code for dll"s (.dll and .so)? Best regards, -- |\/\/| Seweryn Habdank-Wojewódzki \/\/ Professionalism in programming - www.accu.org ---------------------------------------------------- Hospicja w całej Polsce pomagają osobom umierającym i ich bliskim. Ty też możesz pomóc. Przekaż 1% podatku Fundacji Hospicyjnej (KRS 0000201002). http://klik.wp.pl/?adr=www.httz.pl&sid=685

1 0

Regarding the plod_iterator
by ashish deshpande 06 Apr '09

06 Apr '09

Hello all, I am using the plod_iterator() from the Boost Graph Library for generating a scale-free graph. Actually, I need the generated edges to be distinct. However, I'm getting repeated edges in the generated graph. Is there a way to make sure the edges of the generated scale-free graph are distinct. Thanks, Ashish

2 1