November 2007 - Boost-users

Is Anyone going to tell me the way to Build BOOST?
by question boost 14 Nov '07

14 Nov '07

Hi, at boost root directory i do: C:\Program Files\boost> bjam ^More? --build-dir=C:\Program Files\boost \directory_1 ^More? --toolset=gcc stage the error message is the following :could not find main target assuming it's a name of a file to create could not find main target C:\Program Files\boost \directory_1 assuming it's a name of a file to create... note:C:\Program Files\boost is my boost root directory, stage a subdirectory of directory_1 Be very clear THANK YOU, David _________________________________________________________________ Découvrez le blog Messenger Le Meilleur du Web : toutes les vidéos qui buzzent le plus sur Internet ! http://meilleurduweb.spaces.live.com/

6 8

Boost - Build Directory Problem
by chandra_india＠webdunia.com 14 Nov '07

14 Nov '07

2 1

[filesystem] sugges: directory_begin(), directory_end()
by v2cechura 14 Nov '07

14 Nov '07

I would suggest introduction of (template) functions that would return "begin" and "end" directory iterators. I think that it's clearer to see: for_each(directory_begin( dir_path ), directory_end(), do_something()); than for_each(directory_iterator( dir_path ), directory_iterator(), do_somethig()); The functions directory_begin()/directory_end() would mimic the convenient container.begin()/container.end() functions used in STL. I know that it's merely a decoration but so is std::make_pair. It could just be: template <class Path> inline class basic_directory_iterator<Path> directory_begin(const Path& dp) { return basic_directory_iterator<Path>(dp); } template <class Path> inline class basic_directory_iterator<Path> directory_end() { return basic_directory_iterator<Path>(); } Vaclav Cechura

1 0

Boost semaphore with condition variable
by f＠rtfs.org 14 Nov '07

14 Nov '07

Hi! I have a quite simple problem, but I still lack a nice implementation in C++ with boost. My problem is, that I have two threads A and B and thread A needs some information that B will calculate at runtime. Therefore I need some kind of synchronization. So my first attempt was to use a mutex and a condition variable: Thread A: boost::thread::mutex::scoped_lock lock ( mMutex ); mConditionVar.wait ( lock ); Thread B: mConditionVar.notify_one (); This works nice as long as thread A already started waiting on the condition before thread B finished providing it. Otherwise the notify will be lost. So my second attempt was to use a barrier: Init barrier with count 2 Thread A: mBarrier.wait (); Thread B: mBarrier.wait (); This works fine and A will wait until the result is available and it doens't matter if it starts waiting before or after B finished calculaating. But the problem now is that B also needs to wait for A to reach the point in execution. So if I just want to signal thread A from thread B that some point was reached . I don't know how to do it with boost. If I would have a semaphore available I could init it with 0 and thread A tries to decrement it while thread B will increment it once it reached the point. But http://www.boost.org/doc/html/thread/faq.html#id1747096 says: "Semaphore was removed as too error prone. The same effect can be achieved with greater safety by the combination of a mutex and a condition variable." So how exactly is this achived with the mutex and condition variable. I always only come up with a mutex, condition variable and a bool which all three must be shared between both threads. I would greatly appreciate any suggestions! Thanks, Fabian

1 0

Re : RE : Re: compilation trouble with boost::regex
by hallouina-ml＠yahoo.fr 14 Nov '07

14 Nov '07

Thanks for your answer, I think it's clear : So you mean that BOOST_REGEX_MATCH_EXTRA is made for matching for exemple somethingsomething in the string : "blablablasomethingsomethingblablabla". Is it right? In fact I want extract all the link from a webpage. I want to used the regex because this is faster than doing a DOM tree from the webpage. So I want doing the perl equivalent of @matches =~ m/.*(regex_of_the_url_with_href).*/gs g flag because I want extract all of the url. s flag beacause I want that "." match "\n" too. ----- Message d'origine ---- De : John Maddock <john(a)johnmaddock.co.uk> À : boost-users(a)lists.boost.org Envoyé le : Mardi, 13 Novembre 2007, 17h13mn 26s Objet : Re: [Boost-users] RE : Re: compilation trouble with boost::regex hallouina-ml(a)yahoo.fr wrote: > I need Perl style capture like m/(some_regex)/g > > because I need the g flag (to capture more than one time my regex). So > if I understand the documentation (as you can see, my english is far > from beeing perfect...) ; if I want to use the g flag equivalent in my > regex, I need to enable this feature. Oh no you don't :-) The Perl /g flag is something different again: the equivalent functionality in Boost.Regex is regex_iterator, which enumerates one at a time over all the possible matches in a string. Repeated captures come into play when you have say an expression "(something)+" matched against "somethingsomething" in which case in Perl: $0 = "somethingsomething" $1 = "something" With BOOST_REGEX_MATCH_EXTRA enabled then $1 contains a list of all the "something"'s matched by the first capturing group within a single overall match. I hope this is clear, Regards, John Maddock. _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users _____________________________________________________________________________ Ne gardez plus qu'une seule adresse mail ! Copiez vos mails vers Yahoo! Mail

3 2

boost-users [program options, tuple]
by chun ping wang 14 Nov '07

14 Nov '07

Hi i would like to know how to make program options play nicely with tuple. int main( int ac, char* av[] ) { namespace po = boost::program_options; po::options_description desc("Allowed options"); std::vector<BigInteger> key_pair; desc.add_options() ("generate", "generates a pair of key") ("secert", po::value(&key_pair)->multitoken(), "creates the share secret key"); po::variables_map vm; po::store(po::parse_command_line(ac, av, desc), vm); po::notify(vm); try { if (vm.count("generate")) { BigInteger my_prime = encryptcpw::random_prime<BigInteger>(&algocpw::Solovay_Stressen<BigInteger>, 16); std::cout << "my prime: " << my_prime << "\n"; boost::tuple<BigInteger, BigInteger> pare(encryptcpw::dh_init<BigInteger>( my_prime )); std::cout << "pair of generated number: " << pare << "\n"; } if (vm.count("secert")) { const BigInteger p( encryptcpw::random_prime<BigInteger>(&algocpw::Solovay_Stressen<BigInteger>, 16)); std::cout << encryptcpw::dh_secret<BigInteger>( p, key_pair[0], key_pair[1]); } } catch (const char* error) { std::cerr << error << "\n"; } } I would like to replace vector with tuple. boost::tuple<BigInteger, BigInteger> key_pair; for some reason doing that cause the program to crash.. when run DiffieHellman --generate 1371 2397

1 0

template specializing question (boost::static_visitor related)
by jono 14 Nov '07

14 Nov '07

hello all. i have a boost::static_visitor that delegates to a static function to enable partial specialization... /// the templates & so on are declared as so ...... /// /// /// ============================================================= typedef enum {ADD, SUBTRACT, MULTIPLY, DIVIDE, EQUAL, LESS} BinaryOp; template<BinaryOp op, typename ObjectT> Object binary_op(const ObjectT &arg1, const ObjectT &arg2) {return undefined;} template<BinaryOp op, typename ObjectT, typename ObjectU> Object binary_op(const ObjectT &arg1, const ObjectU &arg2) {return undefined;} template<BinaryOp op> class BinaryOpVisitor : public boost::static_visitor<Object> { public: template<typename ObjectT, typename ObjectU> Object operator()( const ObjectT &arg0, const ObjectU &arg1) const { return binary_op<op, ObjectT, ObjectU>(arg0, arg1); } template<typename ObjectT> Object operator ()(const ObjectT &arg0, const ObjectT &arg1) const { return binary_op<op, ObjectT>(arg0, arg1); } }; /// ============================================================= /// /// then i have specialized the binary_op function as follows.... /// /// ============================================================= template<> Object binary_op<ADD>(const Integer& arg0, const Integer& arg1) {return make< Integer >(arg0 + arg1);} template<> Object binary_op<ADD>(const Real& arg0, const Real& arg1) {return make< Real >(arg0 + arg1);} /// ============================================================= /// /// the problem arises when i want to apply an operator to heterogenous but convertible types, eg: Integer + Real /// /// ============================================================= template<> Object binary_op<ADD>(const Integer& arg0, const Real& arg1) {return make< Real >(arg0 + arg1);} template<> Object binary_op<ADD>(const Real& arg0, const Integer& arg1) {return make< Real >(arg0 + arg1);} /// ============================================================= /// /// as shown, i can get the result i want by explicitly specializing each combination of operands, /// but if i add a Complex, then a Rational, then a Modulus and so on, i end up with a combinatorial explosion. /// /// i would like to find a way of doing all this with a single specialization for each pair of heterogenous operands, /// such that they can be accepted in any order... /// any takers? cheers Jono Poff Day One Digital Media Auckland, NZ

1 0

pulling my hair out over Leopard
by Mark Roden 14 Nov '07

14 Nov '07

Hi everyone, I'm trying to get leopard and boost to play nice, so that I can use dicomlib for another app. Apparently, dicom lib requires boost, because it gives 116 error messages if you try to compile in xcode without it. To repro: 1) download dicomlib from http://dicomlib.swri.ca/joomla/index.php?option=com_remository&Itemid=26&fu… 2) put those files into an xcode project 3) watch as they don't compile with 116 errors all related to boost I've tried: 1) including the boost header files in the xcode project. xcode doesn't care unless I change all the < to ", which is painful, time consuming, and frankly, stupid. 2) installing boost at /usr/local/include using ./configure make sudo make install as an administrator. Of course, the boost library is handily titled 'boost-1_34_1' instead of something like 'boost', which means that I have to soft link it using ln -s. Neither the regular install nor the ln option gets around the 115 compile errors. 3) I tried bjam -sTOOLS=darwin and bjam -sTOOLS=darwin -sBUILD=framework, neither of which work, both of which give errors that ld -h is an unrecongized option. I don't know where the linker options are defined, so maybe just getting rid of -h would do it, but that switch probably means something, so just wholesale removing a compiler switch may not be the best move. So here's the real question: How do I make this work? I've lost about 24 hours to this, and I just don't see how it's gonna work. In the meantime, I'm guessing I'm gonna go with http://www.apple.com/downloads/macosx/math_science/quickdicomtoolkit.html but knowing my luck, that will require something similar as well.... Thanks for your help, Mark

5 8

[BGL] Problem while "clearing" a graph
by Olivier Tournaire 13 Nov '07

13 Nov '07

Hi all, i'm trying to clear an adjacency_list. I do the following, as described in the documentation : void MPPRectangleGraph::Clear(unsigned int flags) { bool erase_nodes = false, erase_edges = false, erase_all = false; if ( flags&GraphErase::eERASE_ALL ) erase_all = true; else { if ( flags&GraphErase::eERASE_NODES ) erase_nodes = true; if ( flags&GraphErase::eERASE_EDGES ) erase_edges = true; } if ( erase_nodes || erase_all ) { std::cout << "begin" << std::endl; MPPRectangleGraph::vertex_iterator vi, vi_end, next; boost::tie(vi, vi_end) = vertices(*this); unsigned int i=0; std::cout << num_vertices(*this) << std::endl; //clear_vertex(*vi, *this); //remove_vertex(*vi, *this); for (next = vi; vi != vi_end; vi = next) { std::cout << i++ << std::endl; ++next; clear_vertex(*vi, *this); remove_vertex(*vi, *this); } } std::cout << "end" << std::endl; } The MPPRectangleGraph is simply derived from adjacency_list. Containers are boost:vecS for both vertices and edges. When I have 100 vertices in my graph, I catch an unhandeld exception with catch(...) ("i" is printed only unitl 50), and, when my graph contains only 3 nodes, a std::exception is catched with this message : "vector<T> too long". Can you help ? Regards.

2 3

bind, template member function and thread
by Olivier Tournaire 13 Nov '07

13 Nov '07

Hi all, I would like to use thread with a member function which takes parameters. Here is the code I actually use : shared_ptr<Profile> p2 ( new Profile(origine+(double)i*width_sur_nb_fois_wdir,rect->GetHeightDirection(),MPPParameters_.ProfilesNbPoints_,MPPParameters_.ProfilesDistancePoints_ ) ); shared_ptr<Profile> p3 ( new Profile(origine+(double)(-i)*width_sur_nb_fois_wdir,rect->GetHeightDirection(),MPPParameters_.ProfilesNbPoints_,MPPParameters_.ProfilesDistancePoints_ ) ); p2->ComputeRadiometries(image_,BILINEAR); p3->ComputeRadiometries(image_,BILINEAR); ComputeRadiometries is a template member function : template <class T> bool ComputeRadiometries( boost::shared_ptr< TTImage <T> > image , const eInterpolatorType type_interpolator=BICUBIC ) This compiles fine : boost::function<bool (Profile*, boost::shared_ptr< TTImage <unsigned char> >, eInterpolatorType)> f; But, this does not : f = &Profile::ComputeRadiometries; I would like to use boost::bind in order to call ComputeRadiometries with a thread. Can you help ? Regards -- Le temps des cerises reviendra. Dans l'immédiat, c'est le temps des noyaux. Courage.

3 5