October 2008 - Boost-users - lists.boost.org

Problem linking boost filesystem 1_36_0 on ubuntu gcc 4.2
by Kevin Stevens 30 Oct '08

30 Oct '08

Hello, when I am trying to link boost::filesystem into my program I am receiving link errors. I am trying to convert from boost 1_34_1 to 1_36_0- previously this built against 1_34_1 without any issue. When I link I get the following error: /home/kstevens/code/CrossingEngine/../lib/libtypes.a(Config_Constants.o): In function `boost::enable_if<boost::filesystem::is_basic_path<boost::filesystem::basic_path<std::basic_string<char, std::char_traits<char>, std::allocator<char> >, boost::filesystem::path_traits> >, bool>::type boost::filesystem::exists<boost::filesystem::basic_path<std::basic_string<char, std::char_traits<char>, std::allocator<char> >, boost::filesystem::path_traits> >(boost::filesystem::basic_path<std::basic_string<char, std::char_traits<char>, std::allocator<char> >, boost::filesystem::path_traits> const&)': /usr/include/boost/filesystem/operations.hpp:279: undefined reference to `boost::filesystem::detail::status_api(std::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, int&)' I did an objdump -CT on libboost_filesystem-gcc42-mt-1_36.so and got the following two functions as output but not the one the linker appears to be looking for: 000000000000c4e0 g DF .text 00000000000000e3 Base boost::filesystem::detail::status_api(std::string const&, boost::system::error_code&) 000000000000c720 g DF .text 00000000000000e3 Base boost::filesystem::detail::symlink_status_api(std::string const&, boost::system::error_code&) Could this be a bad build? Am I missing something- perhaps I need to link another file in? I built 1_36_0 myself using a default configuration other than telling it where to put the output libs. I have ensured that I am linking against and including headers from 1_36_0, and have also made a clean build. Is there anything else I should try? Thank you, -Kevin This is my code snippet that uses boost::filesystem: //Create Directory if it does not already exist. if(!boost::filesystem::exists(reportFilePath)) { cout << "output path: " << reportFilePath << " did not exist, which may indicate a different configuration than was intended, creating directory now." << endl; try { boost::filesystem::path path(reportFilePath); boost::filesystem::create_directory(path); } catch(boost::filesystem::filesystem_error e) { cout << "Error: " << e.what() << endl; } }

1 0

non-greedy regex
by Jeff Anderson 30 Oct '08

30 Oct '08

Hello, I'm using boost's regex for a school project (they encourage us to use boost) Boost 1.34 is what is available on the school computers (unless fedora 9 promotes a newer version out of testing before Tuesday.) I'm trying to use a non-greedy quantifier in a regex. Here is an example from my code: string input=string("/blog/2008/../"); string expr=string("/.*?/\\.\\./"); string replacement=string("/"); std::ostringstream stream(std::ios::out | std::ios::binary); std::ostream_iterator<char, char> stream_iterator(stream); boost::regex pattern(expr, boost::regex::perl); boost::regex_replace(stream_iterator, input.begin(), input.end(), pattern, replacement, boost::match_default | boost::format_all); string result=stream.str(); The expected output is "/blog/". I am getting "/". I've played with different arguments to the boost::regex constructor, but I can't seem to get it to output a correct result. I did notice that in the notes for the boost regex library in 1.34 that there was a bug fixed related to non-greedy quantifiers, so I'm fairly confident that I just have a problem with my own code. A tarball can be provided with a "ready-to-compile" example snippet and test case. Then again, it could just as likely be a problem with the regular expression itself, though I checked, re-checked, and double checked that this is the correct syntax for the perl mode in the boost regex library. Thanks for any input! Jeff Anderson

2 2

boost + stlport : Failed to specialize function template
by gatotkaca 30 Oct '08

30 Oct '08

Hello! I'm trying to compile following code (msvc8 + boost1.36.0 + stlport5.1.7) #include <boost/range/size.hpp> #include <iostream> #include <vector> template< class InputRng, class T > inline T accumulate( const InputRng& r, T init ) { return std::accumulate( boost::begin(r), boost::end(r), init ); } template <typename Range> typename boost::range_iterator<Range>::type::value_type mean(const Range &range, typename boost::range_iterator<Range>::type::value_type init= #ifndef BOOST_MSVC typename #endif boost::range_iterator<Range>::type::value_type()) { return typename boost::range_iterator<Range>::type::value_type( boost::accumulate(range, init)/boost::size(range)); } int main(){ std::cout << "test" << std::endl; std::vector<double> data; data.push_back(1); data.push_back(2); data.push_back(3); double m = mean(data); std::cout << m << std::endl; return 0; } It keeps giving me: error C2893: Failed to specialize function template 'boost::range_iterator<C>::type::value_type mean(const Range &,boost::range_iterator<C>::type::value_type)' ..\my works\misc\mean_range_test\testapp.cpp 35 BUT, if I remove the dependency to the STLport (and use MSVC's STL), it works fine! Unfortunately, I need this STLport to work on. I really appreciate for any helps. Thank you! Best, Dio -- View this message in context: http://www.nabble.com/boost-%2B-stlport-%3A-Failed-to-specialize-function-t… Sent from the Boost - Users mailing list archive at Nabble.com.

2 1

Installing Boost.MPI [bjam]
by Raymond Wan 30 Oct '08

30 Oct '08

Hi all, I need some help with building Boost.MPI. My system is a Debian testing (lenny) system and I'm following the instructions at [lenny uses 1.35 and not 1.36]: http://www.boost.org/doc/libs/1_35_0/doc/html/mpi/getting_started.html I'm using Open MPI and I can compile and run the test implementation fine (mpi-test.cpp). So, I moved to the next step on configuring and building. My bjam is 3.1.16. I copied user-config.jam to my home directory, with the line "using mpi ;" added at the end. Then, I ran "bjam --with-mpi" in the "top-level Boost directory" [*] and I get this error: error: no Jamfile in current directory found, and no target references specified. Firstly, I'm not sure if I have everything installed... But I also don't know what the "top-level Boost directory" is. I've tried these locations: /usr/share/boost-build/ /usr/share/boost-build/build/ /usr/share/boost-build/tools/ [it had an mpi.jam file in it] /usr/include/boost/ And also replaced "using mpi ;" with "using mpi : /usr/bin/mpic++ ;". I'm currently trying this as root user, so I don't think root privileges is the problem. (bjam came with a "Hello World" example and that compiled fine; so I don't think it's something wrong with bjam.) Any help would be appreciated... Thanks! Ray

2 4

bind for_each and a range of functors
by Michael Marcin 30 Oct '08

30 Oct '08

Let's say I have a std::list of a functor with a member function virtual operator()( int, int ) Is there an easy way to call each function in this list using for_each? Intuitively I'd expect to be able to do something like: void call_each( std::list<functor>& l ) { std::for_each( l.begin(), l.end(), boost::bind(_1,5,20) ); } Thanks -- Michael Marcin

4 5

Question about Protocols and boost::asio::ip::tcp::iostream
by Charles Milutinovic 30 Oct '08

30 Oct '08

I am trying to put together a program that needs a very simple networking component, esentially just dumping a lot of data from iostreams into a socket, unformatted, and tcp::iostream looked perfect for it. I have a question about its usage though. Every example I've seen looks something like tcp::iostream s("server", "protocol"). So far I've seen "http" and "daytime" as protocols. I am wondering exactly where these constants are defined. Also, for my purposes, how hard would it be to extend a previously defined protocol (i.e. telnet) since all I really need to do is connect to a port of my choosing and dump data.

2 1

tellg() not working for boost::iostreams::filtering_istream
by Roland Bock 30 Oct '08

30 Oct '08

Hi, tellg() member function is not working for boost::iostreams::filtering_istream (see attached example). Is this a bug or a necessity? Regards, Roland

1 0

[serialization] Assertion failure with shared_ptr cycles
by Ult Mundane 30 Oct '08

30 Oct '08

Hi, I'm hitting an assert when I deserialize cycles of shared_ptr<>s. I have the following class hierarchy: Base <-- Derived Other Derived holds a shared_ptr<Other>, and Other holds a shared_ptr<Base>. I have an instance of Derived that points to an instance of Other, which in turn points back to the Derived instance (as a Base pointer). The serialization goes as expected. On deserialization, however, the Base::serialize() and Derived::serialize() are called repeatedly, crashing after the third time -- Other::serialize() is never called. This problem only happens when the cycle includes a pointer to a base class -- a straightforward a -> b -> a cycle works as expected. Has anyone else experienced this or found a workaround? It happens in both boost 1.36 and 1.37.0 beta1. I'm using gcc (3.3.5, 3.4.6, and 4.3.2). Thanks, Ult P.S. Attached is the source code illustrating the above example. Here is the output: Serializing... Derived Base Base done Other Other done Derived done Deserializing... Derived Base Base done Derived Base Base done Derived cycles: ./boost_1_37_0_beta1/libs/serialization/src/basic_iarchive.cpp:436: const boost::archive::detail::basic_pointer_iserializer* boost::archive::detail::basic_iarchive_impl::load_pointer(boost::archive::detail::basic_iarchive&, void*&, const boost::archive::detail::basic_pointer_iserializer*, const boost::archive::detail::basic_pointer_iserializer* (*)(const boost::serialization::extended_type_info&)): Assertion `new_cid == cid' failed. Aborted #include <fstream> #include <iostream> #include <boost/archive/text_oarchive.hpp> #include <boost/archive/text_iarchive.hpp> #include <boost/serialization/shared_ptr.hpp> #include <boost/serialization/is_abstract.hpp> #include <boost/serialization/export.hpp> // Forward declarations class Base; class Derived; class Other; typedef boost::shared_ptr<Base> BaseRef; typedef boost::shared_ptr<Derived> DerivedRef; typedef boost::shared_ptr<Other> OtherRef; // For debug output struct Debug { Debug(const char * name) : mName(name) { std::cout << std::string(++sIndentLevel, ' ') << mName << std::endl; } ~Debug() { std::cout << std::string(sIndentLevel--, ' ') << mName << " done" << std::endl; } static int sIndentLevel; char const * mName; }; int Debug::sIndentLevel = 0; // Base class Base { public: virtual ~Base() { } template<class Archive> void serialize(Archive & ar, const unsigned int) { Debug d("Base"); } }; // Derived class Derived : public Base { public: void setOther(OtherRef other) { mOther = other; } template<class Archive> void serialize(Archive & ar, const unsigned int) { Debug d("Derived"); ar & boost::serialization::base_object<Base>(*this); // This line is executed three times before a crash on deserialzation ar & mOther; // Code after here is never executed on deserialization } OtherRef mOther; }; // Other class Other { public: void setBase(BaseRef base) { mBase = base; } template<class Archive> void serialize(Archive & ar, const unsigned int) { Debug d("Other"); ar & mBase; } BaseRef mBase; }; BOOST_CLASS_EXPORT(Derived); // main int main(int /* argc */, char * /* argv */[]) { char const * fname = "cycles.text_archive"; std::cout << std::endl << "Serializing..." << std::endl << std::endl; { DerivedRef derived(new Derived()); OtherRef other(new Other()); derived->setOther(other); other->setBase(derived); std::ofstream ofs(fname, std::ios::out | std::ios::binary); boost::archive::text_oarchive oa(ofs); oa & derived; } std::cout << std::endl << "Deserializing..." << std::endl << std::endl; { Derived derivedCopy; std::ifstream ifs(fname, std::ios::in | std::ios::binary); boost::archive::text_iarchive ia(ifs); ia & derivedCopy; } }

2 1

Bug or feature? date_time::microseconds limitations?
by athor 30 Oct '08

30 Oct '08

The boost::date_time::microseconds(...) constructor overflows the int64 size limits in intermediary calculations. The code below will reproduce the error (using VC++ 2008 and boost 1.36). ptime start(boost::gregorian::date(2000, 1, 1)); ptime end(boost::gregorian::date(2000, 5, 1)); time_duration td = end - start; ptime wrongEnd = start + microseconds(td.total_microseconds()); BOOST_CHECK_EQUAL(end, wrongEnd); The microseconds constructor takes an int64 as input and the total_microseconds is well within its range. Inside the date_time::subsecond_duration<time_duration, 1000000> the internal duration is calculated as ss*traits_type::res_adjust()/frac_of_second Since the both res_adjust() and frac_of_second have the value 1000000, the result should be ss*1 but... the intermediary value ss*res_adjust overflows the int64 causing a negative number to be returned. Question is, is this by design or is it a bug? I'm rather new to boost::date_time and was hoping to replace our current datetime types with it. We are using millisecond timestamped data over long time series (years). I guess the simple solution would be to add parentheses around (res_adjust()/frac_of_second) assuming frac_of_second always divides res_adjust. However, there are similar issues when adding 200 years worth of seconds which overflows the int32 argument (but that's not of my immediate concern). -- View this message in context: http://www.nabble.com/Bug-or-feature--date_time%3A%3Amicroseconds-limitatio… Sent from the Boost - Users mailing list archive at Nabble.com.

1 0

boost.python / bjam
by Mark Williams 30 Oct '08

30 Oct '08

Hi, is it possible to build boost.python such that the python version gets embedded in the .so name along with the compiler name, version, multithread flags, etc? For example, "libboost_python25-gcc41-mt-1_35.so" instead of "libboost_python-gcc41-mt-1_35.so" Alternatively, is it possible to build it without the embedded SONAME so that it can be renamed afterwards? Thanks in advance, Mark

2 1