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

[asio] Problem with async_read/async_read_some not reacting to incoming data
by Stig Sandø 19 Mar '09

19 Mar '09

Hei, We're rewriting a client with boost asio but have run into some problems when stresstesting the client. The client is fetching textual and graphics from a server with one connection that is open at all time. When the client is getting large amounts of graphics data it will after awhile suddenly stop receiving data, and eventually it will hit our timeouts and the last call sent is an async_read/async_read_some. We are keeping the server well-fed with requests so there should be graphics forthcoming without pause. The problem has been seen on win32, linux and darwin when testing on a gigabit net fetching raw 1080i graphics (4M for each field), and is most frequent on darwin. This is naturally an absolute show-stopper for us. So we are a bit loss what is going wrong and why async_read/async_read_some stops reacting in the middle of the fetch-queue, despite wireshark showing that the data is incoming. When using compression on the data the problem is harder to reproduce, which might suggest a race-condition somewhere. But our code is just using a single thread for io_service and all async-communication is triggered from this io-thread which has a work-object to keep the io_service spinning. We're also making sure there is at most one async_read and one async_write in effect at a time, roughly similar to the chat_client sample. Has anyone seen something similar or have any input on how best to figure out what goes wrong? Are there invariants that says you cannot read and write at the same time? Some symptoms are the same in each test. When we get the last image from the socket the buffersize is zero afterwards, and the next async_read request is to transfer_at_least(1). The async_read never calls the handler for completion of this byte, so Nagle would have kicked in. It is also fairly hard to strip down to a small example using a mock server. I have included some stripped down code below in case that might be helpful spotting something that we cant see. Cheers, Stig Stacktrace during timeout of the asio-thread: (gdb) bt #0 0x91302f66 in kevent () #1 0x003e8010 in boost::asio::detail::kqueue_reactor<false>::run () #2 0x0045f70a in boost::asio::detail::task_io_service<boost::asio::detail::kqueue_reactor<false> >::do_one () #3 0x0045f8c3 in boost::asio::detail::task_io_service<boost::asio::detail::kqueue_reactor<false> >::run () #4 0x0045f972 in boost::asio::io_service::run () #5 0x003f8bbc in vcl::connection_pool::implementation::run () #6 0x003fa2fa in boost::_mfi::mf0<void, vcl::connection_pool::implementation>::operator() () #7 0x003fa356 in boost::_bi::list1<boost::_bi::value<vcl::connection_pool::implementation*> >::operator()<boost::_mfi::mf0<void, vcl::connection_pool::implementation>, boost::_bi::list0> () #8 0x003fa399 in boost::_bi::bind_t<void, boost::_mfi::mf0<void, vcl::connection_pool::implementation>, boost::_bi::list1<boost::_bi::value<vcl::connection_pool::implementation*> > >::operator() () #9 0x003fa62e in boost::detail::thread_data<boost::_bi::bind_t<void, boost::_mfi::mf0<void, vcl::connection_pool::implementation>, boost::_bi::list1<boost::_bi::value<vcl::connection_pool::implementation*> > > >::run () #10 0x003b97ce in thread_proxy () #11 0x913036f5 in _pthread_start () #12 0x913035b2 in thread_start () The connection headerfile: #ifndef VCL_ASIO_CONNECTION_HPP #define VCL_ASIO_CONNECTION_HPP #include <vcl/connection.hpp> #include <queue> #include <boost/asio.hpp> #include <boost/thread.hpp> #include <boost/enable_shared_from_this.hpp> namespace soul { struct logger_base; typedef boost::shared_ptr<logger_base> logger; } using boost::asio::ip::tcp; namespace vcl { class connection_manager; class dispatcher; class protocol; class response; typedef boost::shared_ptr<connection_manager> connection_manager_ptr; typedef boost::weak_ptr<connection_manager> connection_manager_wptr; typedef boost::shared_ptr<dispatcher> dispatcher_ptr; typedef boost::shared_ptr<protocol> protocol_ptr; typedef boost::weak_ptr<protocol> protocol_wptr; typedef boost::shared_ptr<response> response_ptr; typedef std::queue<request_ptr> req_queue; enum read_status { rs_no_valid_data, rs_read_one_block, rs_error_occured }; class asio_connection : public connection, public boost::enable_shared_from_this<asio_connection> { public: asio_connection(const soul::logger &logger, boost::asio::io_service& io_service, connection_manager_wptr cmgr, const dispatcher_ptr &disp, const vbl::host_repr_ptr &host, const protocol_ptr &cp); virtual ~asio_connection(); void handle_resolve(const boost::system::error_code& error, tcp::resolver::iterator endpoint_iterator); void handle_connect(const boost::system::error_code& error, tcp::resolver::iterator endpoint_iterator); virtual void lose_connection(); virtual void start_connecting(); virtual vbl::host_repr_ptr get_host(); virtual bool send(const request_ptr &req); static connection_ptr connect(const std::string &host, unsigned short port, const protocol_ptr &cp); private: void handle_write_request(const boost::system::error_code& error); unsigned handle_capture_request(); read_status handle_text_request(); read_status handle_icon_request(unsigned &next_amount); void start_reading(unsigned bytes=1); void handle_read_info(const boost::system::error_code& error, std::size_t tf); read_status _handle_read_info(unsigned &next_amount); void io_send(const request_ptr &req); bool io_dispatch_response(response_ptr myresponse); bool io_dispatch_error(const boost::system::error_code& error); std::string prepared_request(const request_ptr &msg) const; private: soul::logger local_logger; boost::asio::io_service& m_io_service; connection_manager_wptr m_manager; dispatcher_ptr m_dispatcher; tcp::socket m_socket; vbl::host_repr_ptr m_host; tcp::resolver::iterator m_host_iterator; protocol_wptr wcp; req_queue outgoing_requests; req_queue incoming_requests; boost::asio::streambuf inbuffer; request_ptr currently_reading; }; } #endif /* VCL_ASIO_CONNECTION_HPP */ And the stripped down communication-part of the connection object: #include "connection.hpp" #include "../logging.hpp" #include "../internal_response.hpp" #include <vcl/protocol.hpp> #include <vcl/request.hpp> #include "../manager.hpp" #include <vbl/host.hpp> #include <vbl/exceptions.hpp> #include <vcl/callback.hpp> #include <vcl/dispatcher.hpp> #include <boost/bind.hpp> namespace vcl { asio_connection::asio_connection(const soul::logger &logger, boost::asio::io_service& io_service, connection_manager_wptr cmgr, const dispatcher_ptr &disp, const vbl::host_repr_ptr &host, const protocol_ptr &cp) : local_logger(logger), m_io_service(io_service), m_manager(cmgr), m_dispatcher(disp), m_socket(io_service), m_host(host), wcp(cp) { } asio_connection::~asio_connection() { } void asio_connection::start_connecting() { tcp::resolver resolver(m_io_service); tcp::resolver::query query(m_host->get_hostname(), m_host->get_port_as_string()); boost::system::error_code ec; m_host_iterator = resolver.resolve(query, ec); handle_resolve(ec, m_host_iterator); } void asio_connection::handle_resolve(const boost::system::error_code& error, tcp::resolver::iterator endpoint_iterator) { // THIS CAN BE ANY THREAD if (!error) { tcp::endpoint endpoint = *endpoint_iterator; m_socket.async_connect(endpoint, boost::bind(&asio_connection::handle_connect, shared_from_this(), boost::asio::placeholders::error, ++m_host_iterator)); } else { m_io_service.post(boost::bind(&asio_connection::io_dispatch_error, shared_from_this(),error)); } } void asio_connection::handle_connect(const boost::system::error_code& error, tcp::resolver::iterator endpoint_iterator) { if (!error) { if (protocol_ptr p = wcp.lock()) { p->set_connection(shared_from_this()); m_dispatcher->dispatch(make_callback("connection made", boost::bind(&protocol::connection_made, p))); } // Now we start reading start_reading(); } else { io_dispatch_error(error); lose_connection(); } } bool asio_connection::io_dispatch_error(const boost::system::error_code& error) { if (protocol_ptr p = wcp.lock()) { m_dispatcher->dispatch(make_callback("Error occured", boost::bind(&protocol::error, p, error))); return true; } else { return false; } } void asio_connection::lose_connection() { /* vaious cleanup */ } void asio_connection::io_send(const request_ptr &msg) { // if the request expects an answer, push it in the incoming requests queu if (msg->get_type() != noreply_request) { incoming_requests.push(msg); } bool write_in_progress = !outgoing_requests.empty(); outgoing_requests.push(msg); if (!write_in_progress) { std::string data(prepared_request(outgoing_requests.front())); boost::asio::async_write(m_socket, boost::asio::buffer(data.data(), data.size()), boost::bind(&asio_connection::handle_write_request, shared_from_this(), boost::asio::placeholders::error)); } } void asio_connection::handle_write_request(const boost::system::error_code& error) { if (!error) { outgoing_requests.pop(); if (!outgoing_requests.empty()) { std::string data(prepared_request(outgoing_requests.front())); boost::asio::async_write(m_socket, boost::asio::buffer(data.data(), data.size()), boost::bind(&asio_connection::handle_write_request, shared_from_this(), boost::asio::placeholders::error)); } } else { lose_connection(); } } bool asio_connection::send(const request_ptr &msg) { if (!msg.get()) return false; // make sure we do the real stuff on the io-thread m_io_service.post(boost::bind(&asio_connection::io_send, shared_from_this(), msg)); return true; } void asio_connection::start_reading(unsigned bytes) { boost::asio::async_read(m_socket, inbuffer, boost::asio::transfer_at_least(bytes), boost::bind(&asio_connection::handle_read_info, shared_from_this(), boost::asio::placeholders::error, boost::asio::placeholders::bytes_transferred)); } bool asio_connection::io_dispatch_response(response_ptr myresponse) { if (protocol_ptr p = wcp.lock()) { // make a callback and see what happens m_dispatcher->dispatch(make_callback("response received", boost::bind(&protocol::dispatch_response, p, myresponse))); return true; } else { return false; } } void asio_connection::handle_read_info(const boost::system::error_code& error, std::size_t tf) { if (!error) { unsigned next_amount = 0; read_status rs = _handle_read_info(next_amount); while (rs == rs_read_one_block) rs = _handle_read_info(next_amount); if (next_amount < 2) next_amount = 1; start_reading(next_amount); } else if (error.value() == boost::system::errc::operation_canceled) { // ignore this error for now, resolve it later lose_connection(); } else { lose_connection(); } } }

3 2

range transformed -adaptor incompatible with lambda functions
by Sandeep Gupta 19 Mar '09

19 Mar '09

Hi, I would be nice if range-transformed worked with lambda functions. Is this an intended behaviour. To illustrate the following two statements fail for_each(myvec|boost::adaptors::transformed(arg1+mask), std::cout<<arg1<<' '); for_each(myvec|boost::adaptors::transformed(boost::lambda::_1 + mask), std::cout<<boost::lambda::_1<<' '); But this is fine: for_each(myvec|boost::adaptors::transformed(double_int()), std::cout<<boost::lambda::_1<<' '); where double_int hand created function object with result_type declaration. Thanks sandeep

1 0

Re: [Boost-users] share_ptr reset() question
by 张亚霏 19 Mar '09

19 Mar '09

of cource, no memory leak. check out the source code of method 'reset'. there is a little trick help release the memory, which is similar with Effective STL: item 17: Use "swap trick" to trim excess capacity good luck ------------------ Original ------------------ From: "Ben"<befung(a)gmail.com>; Date: Thu, Mar 19, 2009 08:28 AM To: "boost-users"<boost-users(a)lists.boost.org>; Subject: [Boost-users] share_ptr reset() question Hello, Consider the following: void createSharePtr( boost::shared_ptr<int> value) { value.reset( new int(100)); } main() { boost::shared_ptr<int> y( new int(0)); createSharePtr( y); }; Question: Does the function createSharePtr() cause memory leak? Thanks, Ben _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

1 0

share_ptr reset() question
by Ben 19 Mar '09

19 Mar '09

Hello, Consider the following: void createSharePtr( boost::shared_ptr<int> value) { value.reset( new int(100)); } main() { boost::shared_ptr<int> y( new int(0)); createSharePtr( y); }; Question: Does the function createSharePtr() cause memory leak? Thanks, Ben

2 1

Re: [Boost-users] [graph] Convert bidirectional graphs into directed graphs
by Juan Antonio Farré Basurte 19 Mar '09

19 Mar '09

>> I need to convert any bidirectional graph into an >> adjacency_list<vecS,vecS,directedS>, >> but looks like the copy constructor cannot handle it (needs the same type in input). >> I don't see any simple way to convert any directed graph into an adjacency_list<vecS,vecS,directedS>. >> Is there any more-or-less-simple way to do it that I don't know? May be there is a way to copy graphs constrained not by the concrete type of the graph, >> but only, may be, on graph concepts? >> Thanks, > > There's no direct conversion between *any* bidirectional graph and a directed adjancency_list. If you're talking about any bidirectional adjacency list and a directed one, you shouldn't actually need to convert them. > > As an alternative, you could use copy_graph. > I need to do a copy of the graph before writing it into graphml or graphviz as a workaround for the problem I reported with subject [graph] Looks like there is a bug in some graph-writing function , which in my second message with the same subject I clarified the problem was related with creating dynamic_properties from bundled properties. It doesn't work with const graphs and I don't trust just performing a const_cast. I want to make sure that the input graph to my graph-writing functions is not modified at all. One solution would be just to do Graph *copy=new Graph(graph); where Graph is the graph-type template parameter, but this would require Graph to model Assignable, and I don't want to impose that. Also, copying the graph into a directed (not bidirectional) adjacency_list will save memory. copy_graph is perfect for my needs, as graph-writing boost functions already require the graph to model VertexListGraph (and, in general, the code I'm developing too). Sorry for not remembering about copy_graph and thank you very much :) Juan

1 0

Re: [Boost-users] Boost internal link problem
by Frank J. R. Hanstick 19 Mar '09

19 Mar '09

Frank J. R. Hanstick wrote: > /bin/sh ../libtool --tag=CXX --mode=link g++ -D_THREAD_SAFE -g -O2 > -lboost_system-xgcc40-mt-1_37.dylib -o browser browser.o > libtest-openvrml.la -lboost_unit_test_framework-xgcc40-mt-1_37 > -lboost_filesystem-xgcc40-mt-1_37 > /usr/bin/ld: can't locate file for: -lboost_system-xgcc40- mt-1_37.dylib Shouldn't this be -lboost_system-xgcc40-mt-1_37 (no .dylib) Also, why is it in a different place from the other libs? In Christ, Steven Watanabe >> /usr/bin/ld: can't locate file for: -lboost_system-xgcc40- mt-1_37.dylib >Shouldn't this be -lboost_system-xgcc40-mt-1_37 (no .dylib) There is not an -lboost_system-xgcc40-mt-1_37 input to g++ when this occurs, so, it had to be gotten from somewhere else, either internal to g++ which is unlikely or from libboost_filesystem_xgcc40-mt-1_37. >Also, why is it in a different place from the other libs? All Boost libraries are in /usr/local/lib where they are suppose to be. This is verified by the command: make check LDFLAGS='-lboost_system-xgcc40-mt-1_37' successfully finding libboost_system-xgcc-mt-1_37.dylib. Come think of it, there is also a libboost_xxx-xgcc40-mt set of Boost libraries from an earlier install that gave the exact same error message when libboost_system_xgcc40-mt is not specified which would indicate that somewhere within libboost_filesystem_xgcc40-mt and libboost_filesystem_xgcc40-mt-1_37 is an improper link direction generated by the Boost library build that is overridden by the LDFLAGS input. Frank J. R. Hanstick trog24(a)comcast.net

1 0

Boost internal link problem
by Frank J. R. Hanstick 19 Mar '09

19 Mar '09

Hello, Trying to link to the Boost libraries for openvrml resulted in the following error: /bin/sh ../libtool --tag=CXX --mode=link g++ -D_THREAD_SAFE -g - O2 -lboost_system-xgcc40-mt-1_37.dylib -o browser browser.o libtest- openvrml.la -lboost_unit_test_framework-xgcc40-mt-1_37 - lboost_filesystem-xgcc40-mt-1_37 g++ -D_THREAD_SAFE -g -O2 -o .libs/browser browser.o -Wl,- bind_at_load -lboost_system-xgcc40-mt-1_37.dylib ./.libs/libtest- openvrml.a -L/usr/local/lib -L/opt/local/lib /Users/frank/ FirefoxDownloads/openvrml-0.17.12/src/libopenvrml/.libs/ libopenvrml.dylib -L/opt/local/lib/firefox-2.0.0.20 /opt/local/lib/ libjpeg.dylib /opt/local/lib/libpng12.dylib /opt/local/lib/ libfontconfig.dylib /opt/local/lib/libiconv.dylib /opt/local/lib/ libexpat.dylib /opt/local/lib/libfreetype.dylib -lz -lmozjs -lplds4 - lplc4 -lnspr4 -lboost_thread-xgcc40-mt-1_37 - lboost_unit_test_framework-xgcc40-mt-1_37 -lboost_filesystem-xgcc40- mt-1_37 /usr/bin/ld: can't locate file for: -lboost_system-xgcc40-mt-1_37.dylib collect2: ld returned 1 exit status make[2]: *** [browser] Error 1 make[2]: Leaving directory `/Users/frank/FirefoxDownloads/ openvrml-0.17.12/tests' make[1]: *** [check-am] Error 2 make[1]: Leaving directory `/Users/frank/FirefoxDownloads/ openvrml-0.17.12/tests' make: *** [check-recursive] Error 1 The command sequence that led to the error was: ./configure BOOST_LIB_SUFFIX="-xgcc40-mt-1_37" LDFLAGS="-L/usr/local/ lib -L/opt/local/lib" CPPFLAGS=-I"/usr/local/spidermonkey/include/ -I/ Users/frank/FirefoxDownloads/xulrunner-sdk/sdk/include/ -DXP_UNIX" PKG_CONFIG_PATH="/opt/local/lib/pkgconfig/:/usr/local/lib/pkgconfig:/ sw/lib/pkgconfig" --with-libjs make make check A work around to the error was: make check LDFLAGS='-lboost_system-xgcc40-mt-1_37' The question is why the additional -l was needed to link two Boost libraries in order to overcome a dependency between the two Boost libraries. Frank J. R. Hanstick trog24(a)comcast.net

2 1

pointer to member function of multi_index
by Sandeep Gupta 19 Mar '09

19 Mar '09

Hi, I am sure this is due to some trivial mistake; the statement to create a pointer member function insert of multi_index container fails in the following code: typedef multi_index_container< std::pair<int,int>, indexed_by< ordered_unique<member<pair<int,int>, int, &pair<int, int>::first> > > > mic; std::pair<mic_index::iterator, bool> (mic_index::*fn)(pair<int,int>) = &mic_index::insert; //compile error The error is due to mismatch in signature as given in the error: std::pair<boost::multi_index::detail::bidir_node_iterator<boost::multi_index::detail::ordered_index_node<boost::multi_index::detail::index_node_base<std::pair<int, int>, ... std::pair<boost::multi_index::detail::bidir_node_iterator<boost::multi_index::detail::ordered_index_node<typename SuperMeta::type::node_type> >, bool>... If need be I can post full error message. Appreciate any suggestions. Thanks sandeep

3 12

[boost-thread] error in the docs???
by Ovanes Markarian 19 Mar '09

19 Mar '09

Hello *, I think there is an error in the docs at: http://www.boost.org/doc/libs/1_38_0/doc/html/thread/synchronization.html#t… Member Function of condition_variable class is documented as: template<typename predicate_type> void wait(boost::unique_lock<boost::mutex>& lock, predicate_type pred) Effects: As-if while(!pred()) { wait(lock); } Should be: template<typename predicate_type> void wait(boost::unique_lock<boost::mutex>& lock, predicate_type pred) Effects: As-if while(pred()) //!!!!! no not operator { wait(lock); } If I pass the negated predicate to the wait, my thread blocks forever, whereas passing the predicate works as intended. This applies to the version 1.36, but 1.38 states the same in the docs. Thanks, Ovanes

2 3

Re: [Boost-users] [boost_signals] How can i accelerate the compile process under VC.net?
by boost michael 19 Mar '09

19 Mar '09

Hi all, I'm a newbie of boost. I use boost signals mechanism in my project which is developed in VC.net 2003. The boost libs(ver 1_34_1) have already been installed and built correctly. But i still find that the compile precess of each file of my project will become quite slow if a boost function call is included. The total time spending on the compilation of my project may exceed 20 minutes. And I think that is quite long if compared with the compilation of a project with pure funcitons of VC libs. Can anyone give me any advice on how to accelerate the compilation? Thanks ahead. Best, Michael H.

2 1

reference to ‘_1’ is ambiguous
by Sandeep Gupta 18 Mar '09

18 Mar '09

Hi, This problem has appeared before on the list but in a different context. The following code std::vector<int> myvec; for_each(myvec, std::cout<<_1<<' '); gives the error: Serializer.h:78: error: reference to ‘_1’ is ambiguous /home/sandeep/Computing/boost_1_38_0/boost/lambda/core.hpp:69: error: candidates are: const boost::lambda::placeholder1_type& boost::lambda::<unnamed>::_1 /home/sandeep/Computing/boost_1_38_0/boost/bind/placeholders.hpp:54: error: boost::arg<1> <unnamed>::_1 Serializer.h:78: error: reference to ‘_1’ is ambiguous I believe boost/bind is getting included indirectly via some other library. I have tried several modification but none seem to work as I was quite reluctant to post a third mail on the list. Thanks, sandep

8 14

Problem with 'cannot instantiate abstract class' after switching from BOOST 1.33.0 to 1.34.1 (also VS2005 -> VS2008)
by Tobias Alte 18 Mar '09

18 Mar '09

Hi everyone, After switching from BOOST 1.33.0 to 1.34.1 (and also from VS2005 -> VS2008) we got a couple of problems with "cannot instantiate abstract class" errors. The application compiles fine in the setup VS2005 + BOOST 1.33.0 but suddenly started to throw a couple of "cannot instantiate abstract class" errors when using VS2008 + BOOST 1.34.1 . The errors itself get reported for 'concept_check.hpp' but that header did not change between the two BOOST releases (as far as I know - I think I made a diff with no results). I believe that the issue is most likely related to the boost::shared_ptr's to templated classes in our code. I already solved a few of these errors that arose thanks to slightly different types in the method definitions thanks to VS2008 having more built in types. However, the current "cannot instantiate abstract class" errors are persistently eluding my grasp. And the compiler is not helpful with just reporting the error in 'concept_check.hpp' and not where it really is. Questions: Can anyone of you think of additional (compiler) switches for BOOST (or even VS2008) that could help track down these kind of the issues? Did anyone experience similar problems when switching between BOOST versions (in general)? Thanks a lot for reading this and for any advice. T

2 4

Passing arguments to threads
by r ottmanj 18 Mar '09

18 Mar '09

I am working on an application that takes use of boost::thread. During the initial method of my application I need to pass one argument to the method. As per the api documentation for boost::thread I am using the following: boost::thread dataThread(dparse.dataMain,instUUID); dataThread.join(); However, when I attempt to use this, I get the following exception from my compiler. Is there something that I am missing? I do not see how my thread constructor is any different from the constructor in the api documentation. Any help with this is greatly appreciated. error: no matching function for call to ‘boost::thread::thread(<unresolved overloaded function type>, std::string&)’ /usr/include/boost/thread/thread.hpp:38: note: candidates are: boost::thread::thread(const boost::function0<void, std::allocator<boost::function_base> >&) /usr/include/boost/thread/thread.hpp:37: note: boost::thread::thread() /usr/include/boost/thread/thread.hpp:35: note: boost::thread::thread(const boost::thread&)

4 5

[Lexical Cast] Strange behaviour
by Robert Jones 18 Mar '09

18 Mar '09

I know this is not new, but I've just bumped into it.... This code #include <iostream> #include <boost/lexical_cast.hpp> int main(int argc, char* argv[]) { try { ulong u = boost::lexical_cast<ulong>("-1"); cout<<"Conversion"< } catch(const std::exception&) { cout<<"Exception"< } } Silently returns ULONG_MAX under MSVC7, though not under GCC. It's an accuate reflection of how the underlying C library handles it, but do we want to propogate this strangeness? Rob.

4 4

[asio] io_service threadpool performance
by Rutger ter Borg 18 Mar '09

18 Mar '09

Dear all, I have been testing asio's io_service in a threadpool setup for job dispatching. However, it seems as if adding threads doesn't improve performance; perhaps even the opposite with 1 thread having the best performance. See below for some results of a simple test I'm doing, posting 10 M jobs to the io_service, and starting N threads at io_service::run after that. Timings below are measured on an 8-core machine, I would expect the execution performance to improve (not to get worse) for execution by more threads. Posting to the io_service is done single-threaded, so these timings should remain approx. the same. Please find attached the test program. Is there something I've missed and/or should tweak to get the desired improvement per added thread? Many thanks, Kind regards, Rutger ter Borg Concurrency = 1 Finished posting after: 3.15 Finished execution after: 5.44 Execs / sec: 1e+07/2.29=4.36681e+06 Concurrency = 2 Finished posting after: 2.85 Finished execution after: 5.47 Execs / sec: 1e+07/2.62=3.81679e+06 Concurrency = 3 Finished posting after: 3.15 Finished execution after: 11.65 Execs / sec: 1e+07/8.5=1.17647e+06 Concurrency = 4 Finished posting after: 3.15 Finished execution after: 9.8 Execs / sec: 1e+07/6.65=1.50376e+06 Concurrency = 5 Finished posting after: 3.28 Finished execution after: 12.45 Execs / sec: 1e+07/9.17=1.09051e+06 Concurrency = 6 Finished posting after: 3.29 Finished execution after: 8.84 Execs / sec: 1e+07/5.55=1.8018e+06 Concurrency = 7 Finished posting after: 3.51 Finished execution after: 10.09 Execs / sec: 1e+07/6.58=1.51976e+06 Concurrency = 8 Finished posting after: 3.38 Finished execution after: 12.54 Execs / sec: 1e+07/9.16=1.0917e+06

4 6

Range_Ex and Multi-Index: conflicting mpl::begin and boost::begin
by Sandeep Gupta 18 Mar '09

18 Mar '09

Hi, I am under-impression that for_each construct of range_ex for all container concepts. However when I tried with multi-index it gave conflicting errors. Attached below is the code and corresponding errors. code: typedef multi_index_container< std::pair<value_type,int>, indexed_by< ordered_unique<member<pair<value_type,int>, value_type, &pair<value_type, int>::first> >, ordered_unique<member<pair<value_type,int>, int, &pair<value_type, int>::second> > > > mic; mic serialStore; for_each(serialStore, std::cout<<boost::lambda::_1<<' '); error: /home/sandeep/Computing/boost_1_38_0/boost/mpl/begin_end_fwd.hpp:22: error: ‘template<class Sequence> struct boost::mpl::begin’ is not a function, /home/sandeep/Computing/boost_1_38_0/boost/range/begin.hpp:96: error: conflict with ‘template<class T> typename boost::range_iterator::type boost::begin(T&)’ /home/sandeep/Computing/boost_1_38_0/boost/range_ex/./detail/adl_begin_end.hpp:27: error: in call to ‘begin’ /home/sandeep/Computing/boost_1_38_0/boost/mpl/begin_end_fwd.hpp:23: error: ‘template<class Sequence> struct boost::mpl::end’ is not a function, /home/sandeep/Computing/boost_1_38_0/boost/range/end.hpp:92: error: conflict with ‘template<class T> typename boost::range_iterator::type boost::end(T&)’ /home/sandeep/Computing/boost_1_38_0/boost/range_ex/./detail/adl_begin_end.hpp:35: error: in call to ‘end’ I apologies in advance if this is due to trivial omission. Thanks -sandeep

1 0

[xpressive] is regex_match threadsafe?
by Sergey Skorokhodov 18 Mar '09

18 Mar '09

Hi, Is it safe to call regex_match with the same cregex|sregex object from different threads? Or should each thread use its own cregex|sregex object? TIA -- Sergey Skorokhodov

3 3

[program options] allow unregistered options for config file
by Deane Yang 18 Mar '09

18 Mar '09

Instructions for how to allow unregistered options on the command-line are given here: http://www.boost.org/doc/libs/1_38_0/doc/html/program_options/howto.html#id… Is there any way to allow unregistered options for a configuration file, too?

2 2

[boost-users][python]
by Germán Diago 18 Mar '09

18 Mar '09

Hello. I have a library which exposes the following function: namespace ftr { Player & getPlayer(int num); } and it's exposed to python by: def("getPlayer", &ftr::getPlayer, return_value_policy<reference_existing_object>()); This code used to work, but now I compile it and get this error when using it from the python module, under ubuntu 9.04: Traceback (most recent call last): File "/usr/lib/python2.5/site-packages/apport_python_hook.py", line 77, in apport_excepthook pr['PythonArgs'] = '%r' % sys.argv AttributeError: 'module' object has no attribute 'argv' Original exception was: Traceback (most recent call last): File "<string>", line 1, in <module> Boost.Python.ArgumentError: Python argument types in ftr.getPlayer(int) did not match C++ signature: getPlayer(int) Any help here, please? I don't know why this should happen.

2 1

Re: [Boost-users] [graph] set of edges
by Michael Fawcett 18 Mar '09

18 Mar '09

On Mon, Mar 16, 2009 at 5:30 PM, Michael Fawcett <michael.fawcett(a)gmail.com> wrote: > On Mon, Mar 16, 2009 at 5:15 PM, Juan Antonio Farré Basurte > <jafb(a)tinet.org> wrote: > > Please try not to top post per guidelines. I've lost the attributions > unfortunately, but fixed the formatting a little. > >>> I meant for OutEdgeList, however I went back and tried vecS for >>> EdgeList as well and all my tests still passed and my program output >>> was still correct. Perhaps this was corrected in code but the docs >>> weren't updated? >> >> I don't think so... I'm working with version 1.38 and I had a std::map with >> edge_descriptor keys. >> I passed this map, wrapped as an associative_property_map, and added to a >> dynamic_properties object, to graph-reading functions. >> When I traversed the map contents to do something useful with it, looks like >> the descriptors where invalid. In some cases it just looked to do nothing. >> In other cases I got segmentation fault. When I changed to listS, everything >> worked all right. >> In the documentation it just says that it's "not guarranted" to work. And >> looks to be true. Works in some cases and not in others. > > It sounds like something assumes that those descriptors will not be > invalidated unless removed. Maybe Andrew can chime in? > > I would guess it's working for me by accident since I'm never removing > anything from the graphs I create. I'll make sure to keep my usage to > listS for now. Reposting with subject line changed. --Michael Fawcett

3 6

factorial function
by agascon＠correo.ugr.es 18 Mar '09

18 Mar '09

Hi everyone, I'm new to boost (and not so new but not very experienced to c++) and I'm having problems with functions defined with a template, as factorial (http://www.boost.org/doc/libs/1_38_0/libs/math/doc/sf_and_dist/html/math_to…) My questions is, how do I call the funtion factorial? Why isn't it enough calling factorial(n) where n is an unsigned integer? Does one need to define the function specifying the result type before? If so, how? Thanks in advance. Greetings, Alberto.

2 1

How to post a work into a io_service and wait for it run complete?
by �]邓 18 Mar '09

18 Mar '09

I created a io_service in main thread, for dispatch user request. And i created another io_service in a thread_pool, run some sync work(like database query). After main thread got some user request, it post a query into the io_service running in the thread_pool. After the query has been completed, query thread post a work to the main thread. But the next query cannot be handled until the last result was closed. So query thread need to wait main thread until it'll not using the result any more. How to do it? Now I write some code with mutex and condition_variables. but is there any better choice with out such things?

1 0

correct usage of bind
by Sandeep Gupta 18 Mar '09

18 Mar '09

Hi, Both the statements below bind function buildPair. While the first works the other fails. What could be a possible fix. Thanks in advance. //---- Function to be used by bind ------------------------- template<typename X, typename Y> std::pair<X, Y> buildPair(X x, Y y) { std::pair<X,Y> apair(x,y); return apair; } (std::cout<<buildPair(_1,_2).first<<' ')(1,2); ///--compiles fine--------- (bind(buildPair,_1,_2))(3,5); ///-fails------------ Thanks sandeep

2 1

correct usage of bind (pls ignore last email)
by Sandeep Gupta 18 Mar '09

18 Mar '09

Hi, The code below tries to chain function ft1 (argument is a pair) with function buildPair (returns a pair). template<typename X, typename Y> std::pair<X, Y> buildPair(X x, Y y) { std::pair<X,Y> apair(x,y); return apair; } int main() { typedef boost::function<void (std::pair<int,int>)> FuncType1_t; std::vector<std::pair<int,int> > myvec; FuncType1_t ft1 = bind(&std::vector<std::pair<int,int> > ::push_back, &myvec, _1); typedef boost::function<void(std::pair<int,int>)> FuncType2_t; FuncType2_t ft2 = bind(ft1, bind(buildPair<int,int>,_1,_2)); //compile fails return 0; } This fails. The problem however vanished if I replace the declaration "FuncType3 ft2..." with bind(ft1, bind(buildPair<int,int>,_1,_2))(3,5); So my question is what is correct type of bind(ft1, bind(buildPair<int,int>,_1,_2)); Thanks sandeep

2 2

[parameter] type of function arguments
by Rutger ter Borg 17 Mar '09

17 Mar '09

I'm looking into the possibilities of using Boost.Parameter to introduce extra user-friendliness to a part of the numeric-bindings libraries (in this particular case, bindings for BLAS and LAPACK). Boost.Parameter looks like it is exactly what I need, making it easy to write functions with a couple of required, and many optional parameters. What I couldn't immediately catch from the documentation, is what the types of the passed arguments are of a BOOST_PARAMETER_FUNCTION. E.g., the example in section 2.1.4 says BOOST_PARAMETER_FUNCTION( [snip] ) { // ... body of function goes here... // use graph, visitor, index_map, and color_map } but it is not clear to me is how to easily get to the type of graph, visitor, etc., in the body of the function. Thanks in advance, Kind regards, Rutger ter Borg

2 2