- Boost-users - lists.boost.org

Re: [Boost-users] [WAVE] bug: generic lexing error
by Hartmut Kaiser 28 Apr '06

28 Apr '06

> Andreas Sæbjørnsen wrote: > > > I still experience the same problems as before with regards to this > > memory bug. I am not able to test this with the wave driver > though as > > it compiles with a lot of errors in a lot of header files with gcc > > 3.4.3 on linux, but it shows up in the lexed_tokens example. > > That's puzzling me. Why the Wave driver doesn't compile for you? > > Are you sure to really have the corrected version (I've > replaced all memcpy's in cpp_re.cpp with memmove's, which > should avoid the problem reported by valgrind)? Seems that the public CVS currently is a bit out of sync if compared to the 'real' CVS database. At least ViewCVS shows a state which is 5 weeks old. This would explain your compilation problems. And I don't know where you may get a current CVS version, sorry. Regards Hartmut > > Am I correct to believe that HEAD is the default CVS version for > > boost-cvs? I tested this on a default checkout from the boost-cvs. > > HEAD should be the default. > > Regards Hartmut >

2 3

Multi_array: A "proper" iterator over a N>1 Multi_array?
by Greg Link 28 Apr '06

28 Apr '06

I'm using multi_array types as containers for some things I'm working on, and the number of dimensions can get quite large - often over 7, sometimes as high as 11. I'd like to do things such as initialize these multi_arrays, access all elements within them, and other such 'reasonable' operations. Thing is, the only iterator interface to multi_array returns a reference to the nested containers. Thus, to iterate over all sub- elements, I'm forced to have 7-11 nested for-loops to iterate over the entire thing. Is there an existent means of simply iterating over the entire set of data in the multi_array? Has anyone implemented such a thing, or is such a thing something to be desired within boost::multi_array? It seems like such a iterator would be especially useful if also implemented/supported within subgroups, allowing the user to define sub-ranges of the multi_array, then iterate over each element therein, without having to worry about dimensionality, et cetera. Any thoughts, ideas, or pointers to how to easily (without tons of code) iterate over an 11-dimensioned object is appreciated. In conclusion then, I've got two problems. One is "How can I simplify my each-element access to a high-dimensional multi_array", and the second is "Should there be a direct-to-element iterator in multi_array, and does the documentation adequately address the issue of whether or not an iterator points to elements directly or to a sub- graph?" Thanks for your time and troubles, - Greg Link

3 5

[serialization] std::list compiler error
by Stephan Nies 28 Apr '06

28 Apr '06

Hello, I want to use boost/serialization combined with the MPI (Message Passing Interface) to add parallel computation to a traffic simulation i am working on. I want to serialize a STL container filled with custom Objects to send it to another process. If I don't set "boost::serialization::track_never" for every class i get a compiler error. The Example Code (without any MPI) below creates this error when compiled: g++ test3.cpp -I/usr/include -L/usr/lib -lboost_serialization -o test3 /usr/include/boost/archive/detail/oserializer.hpp: In function void boost::archive::save(Archive&, T&) [with Archive = boost::archive::text_oarchive, T = gps_packet]: /usr/include/boost/archive/basic_text_oarchive.hpp:78: instantiated from void boost::archive::basic_text_oarchive<Archive>::save_override(T&, int) [with T = gps_packet, Archive = boost::archive::text_oarchive] /usr/include/boost/archive/detail/interface_oarchive.hpp:78: instantiated from Archive& boost::archive::detail::interface_oarchive<Archive>::operator<<(T&) [with T = gps_packet, Archive = boost::archive::text_oarchive] test3.cpp:76: instantiated from here /usr/include/boost/archive/detail/oserializer.hpp:567: error: invalid application of sizeof to incomplete type boost::STATIC_ASSERTION_FAILURE<false> make: *** [test3] Error 1 Here is a short example that creates the above error based upon the example in the documentation: ---code start-------- #include <iostream> #include <sstream> #include <boost/archive/text_oarchive.hpp> #include <boost/archive/text_iarchive.hpp> #include <boost/serialization/list.hpp> #include <list> class gps_position { private: friend class boost::serialization::access; template<class Archive> void serialize(Archive & ar, const unsigned int version) { ar & degrees; ar & minutes; ar & seconds; } int degrees; int minutes; float seconds; public: gps_position(){}; gps_position(int d, int m, float s) : degrees(d), minutes(m), seconds(s) {} void pretty_print() { std::cout << "deg:" << degrees ; std::cout << " min:" << minutes ; std::cout << " sec:" << seconds << std::endl; } }; typedef std::list<gps_position> list_gps_position; class gps_packet { private: friend class boost::serialization::access; template<class Archive> void serialize(Archive & ar, const unsigned int version) { ar & gps_list; } public: gps_packet(){} list_gps_position gps_list; }; // only when I uncomment the following // boost::serialization::track_never // statements programm will compile //BOOST_CLASS_IMPLEMENTATION(gps_position,boost::serialization::object_serializable); //BOOST_CLASS_TRACKING(gps_position, boost::serialization::track_never); //BOOST_CLASS_IMPLEMENTATION(gps_packet,boost::serialization::object_serializable); //BOOST_CLASS_TRACKING(gps_packet, boost::serialization::track_never); int main() { // create come pseudo data gps_position g0(1, 2, 3.3); gps_position g1(11, 22, 33.33); gps_packet packet; packet.gps_list.push_back(g0); packet.gps_list.push_back(g1); // print before transfer std::cout << "Before:" << std::endl; packet.gps_list.front().pretty_print(); packet.gps_list.back().pretty_print(); // serialize to chars in order to use // the MPI::CHAR elementary datatype std::ostringstream ost; boost::archive::text_oarchive oa(ost); oa << packet; const char* chars = ost.str().data(); // assume some message passing // through MPI::CHAR to happen here std::cout << "Transfering:" << std::endl; std::cout << chars << std::endl; // reconstruction from chars for further // computation std::istringstream ist(chars); boost::archive::text_iarchive ia(ist); gps_packet packet_new; ia >> packet_new; // print transfer result std::cout << "After:" << std::endl; packet_new.gps_list.front().pretty_print(); packet_new.gps_list.back().pretty_print(); return 0; } ---code end-------- when i uncomment BOOST_CLASS_TRACKING(gps_position, boost::serialization::track_never); BOOST_CLASS_TRACKING(gps_packet, boost::serialization::track_never); i can compile and execute with this output: Before: deg:1 min:2 sec:3.3 deg:11 min:22 sec:33.33 Transfering: 22 serialization::archive 3 0 0 0 0 2 0 0 1 2 3.3 11 22 33.330002 After: deg:1 min:2 sec:3.3 deg:11 min:22 sec:33.33 Further Questions: 1. I would also like to use just list<gps_position> or the typedef, instead of the wrapper class gps_packet which i need to pass something to BOOST_CLASS_TRACKING(gps_packet, boost::serialization::track_never); 2. Is it a good Idea to use something like: BOOST_CLASS_IMPLEMENTATION(gps_position,boost::serialization::object_serializable); BOOST_CLASS_TRACKING(gps_position,boost::serialization::track_never); since I don't want to store but just pass the Information or are there any pitfalls? Thanks, Stephan Nies -- When you see this text i am using the webinterface of my mail provider and i apologize for any commercials they may attach below. Echte DSL-Flatrate dauerhaft für 0,- Euro*! "Feel free" mit GMX DSL! http://www.gmx.net/de/go/dsl

1 0

Visual C++ 8 - Example Crashes with Access Violation
by Brown Gabe 28 Apr '06

28 Apr '06

Hi Everyone, Hopefully this isn't some noob question, but here goes... I'm trying to run boost 1.3.1 in Visual C++ 8 VC8 and I keep getting an access violation at the termination of my thread. I've tried looking to see what the problem is but I can't seem to determine the problem. I'm using the plain example thread.cpp ====================[ EXAMPLE CODE ]======================= #include <boost/thread/thread.hpp> #include <boost/thread/xtime.hpp> #include <iostream> struct thread_alarm { thread_alarm(int secs) : m_secs(secs) { } void operator()() { boost::xtime xt; boost::xtime_get(&xt, boost::TIME_UTC); xt.sec += m_secs; boost::thread::sleep(xt); std::cout << "alarm sounded..." << std::endl; } int m_secs; }; int main(int argc, char* argv[]) { int secs = 5; std::cout << "setting alarm for 5 seconds..." << std::endl; thread_alarm alarm(secs); boost::thread thrd(alarm); thrd.join(); } ====================[ EXAMPLE CODE ]======================= I get the expected out from the application. Upon tracing my code, the exception only occurs when the worker thread exits. Exception occured in threadex.c on line 348. Unhandled exception at 0x7c9012b4 in test.exe: 0xC0000005: Access violation reading location 0xcccccccc. " __try { _endthreadex ( ( (unsigned (__CLR_OR_STD_CALL *)(void *))(((_ptiddata)ptd)->_initaddr) ) ( ((_ptiddata)ptd)->_initarg ) ) ;" Any thoughts on what might cause an error like this??? -- -Gabe

4 7

[boost::any] Newbie Question About Compiler Warning on HPUX with boost::any
by Richard Ward 28 Apr '06

28 Apr '06

Hello, I'm a new user to boost and have a question about a compiler warning I'm seeing. I'm currently using just the boost::any type and on HPUX v11.00 PA with aCC: HP ANSI C++ B3910B A.03.56 I'm seeing the following warning: Warning (anachronism) 823: "/Components/dev/Boost/v1.33.1/boost/type_traits/is_reference.hpp", line 38 # Redundant preprocessing concatenation operation results in two valid preprocessing tokens. Depending on this implementation defined behavior will result in non-portable code. BOOST_TT_AUX_BOOL_TRAIT_DEF1(is_reference,T,false) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Here's the compile line that generated the warning: aCC -AA -DACE_HAS_STANDARD_CPP_LIBRARY=1 -D_RWSTD_MULTI_THREAD -D_REENTRANT -U_INCLUDE__STDC_A1_SOURCE -DACE_HAS_THREADS -D_POSIX_C_SOURCE=199506L -DHPUX_VERS=1100 -DACE_LACKS_PRAGMA_ONCE +W302 +W930 +W849 -mt +DAportable +Z -D_LARGEFILE64_SOURCE=1 +O3 +Osize -DACE_NDEBUG -DACE_USE_RCSID=0 -DACE_HAS_EXCEPTIONS -D__ACE_INLINE__ -D__UNIX -c -o .objects/HP-UX.B.11.00_RISC_aCC.A.03.37/release//LicAgentSvc.o LicAgentSvc.cpp Because the boost::any header is included in one of my headers, I see this warning hundreds of times. While I could just suppress it I prefer to first try to figure out what the issue is. I dislike suppressing warnings I don't understand. Does anyone know what this warning means and/or have any suggestions for what to do about it? Thanks, Richard J. Ward Symantec Corp.

1 0

bjam, 64-bit builds, VC8
by Jarl Lindrud 28 Apr '06

28 Apr '06

Hi! I need to build the boost libraries, and eventually my own application, in 64 bi t mode on Windows, but unfortunately I haven't been able to find any information on how to do so. Does anyone here have any tips? Typically, for a 32 build of boost.thread, for instance, I would type "bjam --with-thread -sTOOLS=vc-8_0 stage", but I don't know what the invocation should be for a 64 bit build. I'm using Visual C++ 2005 (the full version, not "express"). Thanks! Jarl.

1 0

multi_index, best way to do insert_or_replace
by bert hubert 28 Apr '06

28 Apr '06

Hi, I misread (or didn't read) the boost::multi_index_container documentation and assumed 'insert' on a unique index would do 'replace', but of course it doesn't and shouldn't. So, I've made this change: - s_negcache.insert(ne); + + pair<negcache_t::iterator, bool> res=s_negcache.insert(ne); + if(!res.second) { + s_negcache.erase(res.first); + s_negcache.insert(ne); + } But I can't help feeling there is probably a better and prettier way. Do people have any suggestions? In this case, ne is a NegCacheEntry, defined as: struct NegCacheEntry { string d_name; QType d_qtype; string d_qname; uint32_t d_ttd; }; And the index: typedef multi_index_container < NegCacheEntry, indexed_by < ordered_unique< composite_key< NegCacheEntry, member<NegCacheEntry, string, &NegCacheEntry::d_name>, member<NegCacheEntry, QType, &NegCacheEntry::d_qtype> >, composite_key_compare<CIStringCompare, std::less<QType> > >, ordered_non_unique< member<NegCacheEntry, uint32_t, &NegCacheEntry::d_ttd> > > >negcache_t; Thanks! -- http://www.PowerDNS.com Open source, database driven DNS Software http://netherlabs.nl Open and Closed source services

2 2

problem with 1.33.1 dll's
by george 28 Apr '06

28 Apr '06

hello! today I switch to 1.33.1 an I get some problems with dynamic libs: until 1.33 I just link with the name of the library -lboost_date_time -l boost_thread now with 1.33.1 I cant, I get: /usr/lib/gcc/i586-suse-linux/4.0.2/../../../../i586-suse-linux/bin/ld: cannot find -lboost_date_time and etc.... autolinking doesn't work so I link with the whole name -lboost_date_time-gcc-mt-1_33_1 -lboost_date_time-gcc-1_33_1 and it get compiled but when I try to run it, it can't find the libs(.so) ./server: error while loading shared libraries: libboost_thread-gcc-mt-1_33_1.so.1.33.1: cannot open shared object file: No such file or directory ./client: error while loading shared libraries: libboost_date_time-gcc-1_33_1.so.1.33.1: cannot open shared object file: No such file or directory but there are those files in the folder..... should I run any utility to fix this? P.S. I build 1.33.1 with: ./configure --prefix=/usr/local make make install thansks for any help.

2 2

[multi_array] How to use non-default constructor?
by Seweryn Habdank-Wojewódzki 28 Apr '06

28 Apr '06

Hi! I would like to know if in multi_array there is a possibility to initialize multi_array using non-default constructor? When I am using std::vector there is a possibility to do it like that: std::vector<MyObject> my_object_vec ( 2, MyObject( parameters ) ); This vector constructor uses non-default constructor of MyObject. In C++ arrays constructon like that should be done in two steps first is to initialize array (uses default constructor) and then we could assign values to the array using copy constructor, but this is not nice, and in my cases not good solution. Important is that in MyObject class there shouldn't be default constructor, and better not to use copy constructor. Of course there exists non-default constructor like MyObject ( const Parameters & parameters ). Regards. -- |\/\/| Seweryn Habdank-Wojewódzki `\/\/

2 4

[MPL] Loooong vectors
by Jaakko Järvi 28 Apr '06

28 Apr '06

Howdy, Are type containers with hundreds of types (maybe even more) at all realistic regarding compilation resources? Is there a container type for such very large type containers already? Access to elements should be at least logarithmic. Vectors of vectors, or a tree structure built out of vectors, would probably do unless there is something better out there already. Best, Jaakko Järvi

1 0