June 2007 - Boost-users - lists.preview.boost.org

ASIO: multiple threads, single io_service
by Andrej van der Zee 22 Jun '07

22 Jun '07

Hi, I have some (beginners) question about the example server3-porgram that ships with the library. It uses multiple threads with a single io_service. In the method server::run() this happens: std::vector<boost::shared_ptr<boost::thread> > threads; for (std::size_t i = 0; i < thread_pool_size_; ++i) { boost::shared_ptr<boost::thread> thread(new boost::thread(boost::bind(&boost::asio::io_service::run, &io_service_))); threads.push_back(thread); } for (std::size_t i = 0; i < threads.size(); ++i) threads[i]->join(); But when are the threads actually scheduled? Does this happen by the shared io_service? Are the registered callback methods for async_read, async_accept, etc executed in the threads? Also, are there any tradeoffs that I should know off for using the single io_service vs io_service-per-thread approach? Thank you, Andrej ___________________________________________________________ Yahoo! Mail is the world's favourite email. Don't settle for less, sign up for your free account today http://uk.rd.yahoo.com/evt=44106/*http://uk.docs.yahoo.com/mail/winter07.ht…

2 1

[ptr_container] How to transfer ownership from smart_ptr to ptr_container
by Seweryn Habdank-Wojewódzki 22 Jun '07

22 Jun '07

Hi! Is there any smart_ptr and ptr_container that can transfer ownership of pointer, to avoid useless copy-construction and destruction of the object. More less scenario can be like that. 1. smart pointer handles pointer to the huge data. 2. pointer is transfered via signal to the object that has queue of them. 2a. remark not all object are going to the queue, so there is a need to keep them in smart pointer. 3. Queue sends away pointer, or internally destroys (respectively to demand of another signal in the program. More less the simplified draft is shown below. Maybe I have to use vector <shared_ptr <T> > instead of ptr_container? Best regards. #include <boost/ptr_container/ptr_vector.hpp> #include <boost/shared_ptr.hpp> #include <iostream> class Foo { public: Foo (int const i) : i_ (i), j_() { ++counter; ++crcounter; std::cout << "Foo (int)" << '\t'; print_counter(); } Foo (Foo const & f) : i_ (f.i_), j_() { ++counter; ++crcounter; std::cout << "Foo (Foo)" << '\t'; print_counter(); } ~Foo() { --counter; std::cout << "~Foo () " << '\t'; print_counter(); } void print (std::ostream & os) const { os << i_; } static int counter; static int crcounter; static void print_counter() { std::cout << "No. Foo = " << counter << " No. created = " << crcounter << '\n'; } private: int i_; int j_[10000]; }; int Foo::counter = 0; int Foo::crcounter = 0; std::ostream & operator<< (std::ostream & os, Foo const & f) { f.print (os); return os; } int main() { using namespace boost; using std::cout; ptr_vector<Foo> pvf; Foo * pf = new Foo(7); cout << *pf << '\n'; pvf.push_back (pf); pf = new Foo(9); cout << *pf << '\n'; pvf.push_back (pf); shared_ptr<Foo> spf (new Foo(11)); cout << *spf << '\n'; pvf.push_back (new Foo (*spf)); spf = shared_ptr <Foo> (new Foo(13)); cout << *spf << '\n'; pvf.push_back (new Foo (*spf)); cout << pvf[0] << '\n'; cout << pvf[1] << '\n'; cout << pvf[2] << '\n'; cout << pvf[3] << '\n'; pf = new Foo(pvf[0]); pvf.erase(pvf.begin()); cout << *pf << '\n'; spf = shared_ptr <Foo> (new Foo (pvf[2])); cout << *spf << '\n'; delete pf; } -- |\/\/| Seweryn Habdank-Wojewódzki \/\/

3 3

(no subject)
by david bautista garcia 22 Jun '07

22 Jun '07

hello: I don't understand this code, anybody can explain to me what does ":" and "" typedef boost::tokenizer<boost::char_separator<char> > tokenizer; boost::char_separator<char> sep(":", "", boost::keep_empty_tokens); tokenizer tokens(str, sep); tokenizer::iterator tok_iter = tokens.begin(); //cout << "TOKEN " << (*tok_iter) << endl; if de input is : TGTBTS:NAME=TGTBTS:42,BSIC=4-6,BCCHFREQ=784,SYSID=DCS1800,MSTXPMAXGSM=<NULL>,MSTXPMAXDCS=0,MSTXPMAXPCS=<NULL>,CELLGLID="214"-"03"-6004-1519; TGTPTPPKF:NAME=TGTBTS:42/TGTPTPPKF:0,RACODE=137,RACOL=0,GRXLAMI=6,GMSTXPMAC=15; what's de output? thanks -------------------------------------------------------------------------------------------- ¿Ya sabes con quién vas? ¿Héroe o Villano? Prepárate para la batalla final en http://friki.ozu.es -------------------------------------------------------------------------------------------- Correo enviado desde http://www.ozu.es

1 0

Need help understanding boost::operators::dividable
by Scott, Steven 22 Jun '07

22 Jun '07

I'm sure the answer to this is out there somewhere, but I'm having trouble finding it. (wearing flame-retardant clothes...) I want to use boost::operators::field_operators to generate all the mathematical operators to allow Vector/scalar arithmetic. Everything works fine except for dividing scalars by vectors. Class Vector looks like this: class Vector : public std::vector<double>, boost::field_operators<Vector, boost::field_operators<Vector, double> > { // required operator+=, -=, *=, /= defined here. }; Now if I write: double x(3.0); Vector v(5), w; v[0] = 1; v[1] = 2; v[2] = 3; v[3] = 4; v[4]=5; w = 3+v; // works fine w = v+3; // works fine ... // subtract and multiply work on both sides w = v/x; // works fine w = x/v; // fails... I think the failure happens because of the way operator/ defined for mixed types something like this: T operator/(U u,T t) { T ans(u); ans/=t; return ans; } which is not what I need. Is there a way I can change the way operator/(double,Vector) works in this one case, but keep the really convenient definitions generated by field_operators otherwise? The information contained in this e-mail is confidential and/or proprietary to Capital One and/or its affiliates. The information transmitted herewith is intended only for use by the individual or entity to which it is addressed. If the reader of this message is not the intended recipient, you are hereby notified that any review, retransmission, dissemination, distribution, copying or other use of, or taking of any action in reliance upon this information is strictly prohibited. If you have received this communication in error, please contact the sender and delete the material from your computer.

2 1

Fwd: [OT] [metaprogramming] Importance of order of delarations/definitions.
by Seweryn Habdank-Wojewódzki 22 Jun '07

22 Jun '07

Hi! Could anyone tell me if there is any example where in e.g. metaprogramming there is a need of strict order of declarations and definitions of templates? Theoretically there should not be a problem, but maybe there are some tricks where order influence on fitting types and/or for lookup of specializations. Generally the question is related to the fact, that GCC force user to put typedefs and declaration/definitions in proper order before any usage of types. And I am looking for the example which points why we can NOT mix all - when the order is important. Regards. -- |\/\/| Seweryn Habdank-Wojewódzki \/\/

2 1

building boost with only one set of libraries
by frederic.bron＠alcan.com 22 Jun '07

22 Jun '07

I have spent so much time to find how to build boost with only one combination of release/debug, multithreading, static/shared... that I want to record the method in the archive for other potential users. By default boost is build 16 times which corresponds to all combinations of: - release/debug - multithreading on/off - static/shared link - runtime static/shared link It takes ages to compile and we get roughly 1 giga-bytes of data. In general you need only one combination of these options but it is not very well documented how to set these options. Here is a simple mean: unix like environnement: $ configure --prefix=path_to_the_intallation_directory edit Makefile: BJAM_CONFIG=link=static threading=multi runtime-link=shared --layout=system LIBS=release $ make install nocygwin environnement (build cygwin independant library from cygwin, can be used in CodeBlocks or with g++ -no-cygwin): build bjam from cygwin (boost-jam-3.1.14.tgz): $ chmod 755 build.bat; ./build.bat gcc-nocygwin build boost: $ full_installation_path_of_bjam/bjam link=static threading=multi runtime-link=shared --toolset=gcc-nocygwin --prefix="windows_path_to_the_intallation_directory" --layout=system release install F. Bron Avis : Ce message et toute pièce jointe sont la propriété d'Alcan et sont destinés seulement aux personnes ou à l'entité à qui le message est adressé. Si vous avez reçu ce message par erreur, veuillez le détruire et en aviser l'expéditeur par courriel. Si vous n'êtes pas le destinataire du message, vous n'êtes pas autorisé à utiliser, à copier ou à divulguer le contenu du message ou ses pièces jointes en tout ou en partie. Notice: This message and any attachments are the property of Alcan and are intended solely for the named recipients or entity to whom this message is addressed. If you have received this message in error please inform the sender via e-mail and destroy the message. If you are not the intended recipient you are not allowed to use, copy or disclose the contents or attachments in whole or in part.

3 2

[signals] How to connect method with more arguments (pointers) to signal
by Seweryn Habdank-Wojewódzki 21 Jun '07

21 Jun '07

Hi How to connect method with more arguments (pointers) to signal using e.g. boost::bind? The code below works, but if I eliminate Data_type class from desing, and try to work directly on _two_ pointers I have a problem with connection methods to signal slots. Could anyone tell me how to do that? Best regards. #include <boost/bind.hpp> #include <boost/signals.hpp> #include <iostream> #include <utility> // some class class Test { public: explicit Test(int x) : x_ (x) {} int const x() const { return x_; } operator int() { return x_; } private: int x_; }; // class wrap references class Data_type { public: Data_type(double & d, Test & t) : d_ (d), t_ (t) {} double & d() const { return d_; } Test & t() const { return t_; } private: double & d_; Test & t_; }; // class react on signals class some_class { public: bool some_function(Data_type & dt) { bool test = (dt.d() > 3.14); std::cout << std::boolalpha << "Check " << dt.d() << " > 3.14 " << test << " and " << dt.t().x() << '\n'; return test; } bool another_function(Data_type & dt) { bool test = (dt.d() > 3.14); std::cout << std::boolalpha << "Check " << dt.d() << " > 3.14 " << test << " and " << dt.t().x() << '\n'; return test; } }; // signal source class Caller { public: Caller() : sig0(), sig1() {} ~Caller(){} typedef boost::signal <bool (Data_type & dt)> signal_type; typedef signal_type::slot_type slot_type; signal_type sig0; signal_type sig1; void emit (double & d, Test & t) { // conversion from 2 parameters to 1 Data_type tmp (d, t); sig0 (tmp); sig1 (tmp); } boost::signals::connection connect_to_sig0 (slot_type const & s) { return sig0.connect (s); } boost::signals::connection connect_to_sig1 (slot_type const & s) { return sig1.connect (s); } }; int main() { using std::cout; using namespace boost; some_class sc; Caller c; // if I will eliminate Data_type class // and change methods that they could take 2 arguments // then bind can not meet requirements of slot_type c.connect_to_sig0 (bind (&some_class::some_function, &sc, _1)); c.connect_to_sig1 (bind (&some_class::another_function, &sc, _1)); Test * t = new Test (4); double * d = new double (3.14); c.emit (*d, *t); } -- |\/\/| Seweryn Habdank-Wojewódzki \/\/

1 0

Compile failure with fairly simple use of multi-index / hashing on Solaris with CC 5.8 (boost 1.34.0)
by Benson Margulies 21 Jun '07

21 Jun '07

I've a relatively basic example of using the multi-index capability that fails to compile on Studio 11 (CC 5.8) on Sparc. I submitted it to the tracker before spotting the injunction to talk to the mailing list first. Can anyone offer any relief here?

5 23

pool allocator size 0
by Sebastian Pfützner 21 Jun '07

21 Jun '07

Hello, the following program crashes in the last line (deallocate) with VC8: #include "boost/pool/pool_alloc.hpp" int main() { boost::pool_allocator<int> a; int* i = a.allocate(0); a.deallocate(i, 0); } I didn't find any annotations about restrictions regarding the allowed size, neither in the c++ standard nor in the boost docs. I thought, calling allocate is the same as calling array new with 0 (gives back a pointer != 0 which must be array deleted). Is the boost pool allocator not fully standard compliant? -- Sebastian Pfützner s.pfuetzner(a)onlinehome.de ICQ-ID: 39965036

1 0

'Explanation' of boostrapping problem
by Benson Margulies 21 Jun '07

21 Jun '07

There is internal dissention as to whether the toolset name is 'sun' or 'sunpro'. Bjam wants 'sun'. regression.py and the configure script want 'sunpro', and the configure script leaves 'sunpro' in user-config.jam with the unpleasant consequences I reported.

1 0