September 2010 - Boost-users - lists.preview.boost.org

Boost.math statistical distributions - testing inverse gamma and inverse chi squared and Bayesan examples - help wanted.
by Paul A. Bristow 16 Sep '10

16 Sep '10

I am working on adding (by popular request;-) a few more statistical distributions inverse gamma and inverse chi squared are close to being done. But I would be most grateful for help with 1 A few spot test values to check that I haven't made mistakes. Ideally I'd like some very accurate values from the ultimate reference Wolfram Mathematica, or perhaps an R add-on (not in main release?). If anyone can spare a moment to evaluate a few values for me please email me privately. 2 I have some suggestions for an example of using these distribution(s), but would like others. (We like to provide a real-life example of statistics in action in our documentation). Do any you followers of the Reverend Bayes have any suggestions for examples, preferably fully worked through from raw data? Thanks Paul PS If there are other distributions for which there are 'popular requests', now is the moment to ask ;-) --- Paul A. Bristow, Prizet Farmhouse, Kendal LA8 8AB UK +44 1539 561830 07714330204 pbristow(a)hetp.u-net.com

1 0

[asio] An error occured in asio::win_mutex.hpp
by Yasuhiko Yoshimura (informax.co.jp) 16 Sep '10

16 Sep '10

Hi ALL, I have encountered this error, and would like to program a win32 FTP client with asio lib. In asio examples I reffenrenced at the 'Services(logger program)'. Could you help me ? I appriciate any suggestion. Thank all, Yasuhiko Yoshimura (Japan) An accsess violence occured at the below ===>> point. win_mutex.hpp:line 125: ------------------------------------------------------------------------ Then variable values are: &crit_section_ 0xcdcdcdcd this 0xcdcdcdcd (boost::noncopyable_::noncopyable) // Locking must be performed in a separate function to lock() since the // compiler does not support the use of structured exceptions and C++ // exceptions in the same function. int do_lock() { __try { ===>> ::EnterCriticalSection(&crit_section_); } __except(GetExceptionCode() == STATUS_INVALID_HANDLE || GetExceptionCode() == STATUS_NO_MEMORY ? EXCEPTION_EXECUTE_HANDLER : EXCEPTION_CONTINUE_SEARCH) { if (GetExceptionCode() == STATUS_NO_MEMORY) return ERROR_OUTOFMEMORY; return ERROR_INVALID_HANDLE; } return 0; } ------------------------------------------------------------------------ Here are my codes below. And error point was ===>> ------------------------------------------------------------------------ using boost::asio::ip::tcp; class ftp_client_dtp : public boost::enable_shared_from_this<ftp_client_dtp> { public: typedef boost::shared_ptr<ftp_client_dtp> pointer; static pointer create(boost::asio::io_service& io_service) { return pointer(new ftp_client_dtp(io_service)); } tcp::socket& socket() { return *work_socket_; } ... void start() { work_thread_.reset(new boost::thread( boost::bind(&boost::asio::io_service::run, &work_io_service_))); // Start an asynchronous resolve to translate the server and service names // into a list of endpoints. tcp::resolver::query query(sip_, sport_); work_resolver_->async_resolve(query, boost::bind(&ftp_client_dtp::handle_resolve, shared_from_this(), boost::asio::placeholders::error, boost::asio::placeholders::iterator)); } void close() { work_io_service_.post(boost::bind(&ftp_client_dtp::do_close, shared_from_this())); } ftp_client_dtp(boost::asio::io_service& io_service) : work_io_service_(), work_resolver_(new boost::asio::ip::tcp::resolver(work_io_service_)), work_socket_(new boost::asio::ip::tcp::socket(work_io_service_)) ===>> { } ~ftp_client_dtp() { work_.reset(); if (work_thread_) work_thread_->join(); } private: std::ofstream f; void handle_resolve(const boost::system::error_code& err, tcp::resolver::iterator endpoint_iterator) { ... } void handle_connect(const boost::system::error_code& err, tcp::resolver::iterator endpoint_iterator) { ... } void handle_read_content(const boost::system::error_code& err) { ... } void do_close() { work_socket_->close(); } std::string sip_, sport_, scurrentpath_, sfilename_; boost::asio::streambuf request_; boost::asio::streambuf response_; boost::scoped_ptr<boost::asio::ip::tcp::resolver> work_resolver_; boost::scoped_ptr<boost::asio::ip::tcp::socket> work_socket_; boost::asio::io_service work_io_service_; boost::scoped_ptr<boost::asio::io_service::work> work_; boost::scoped_ptr<boost::thread> work_thread_; }; ftp_client_dtp::pointer new_connection_; new_connection_ = ftp_client_dtp::create(io_service_); new_connection_->start(); ------------------------------------------------------------------------ Thank all, Yasuhiko Yoshimura (Japan)

2 3

upgrade_to_unique_lock doesn't compile with vc2010
by Braden McDaniel 16 Sep '10

16 Sep '10

Using Boost 1.43, this program: # include <boost/thread.hpp> int main() { boost::shared_mutex mutex; boost::upgrade_lock<boost::shared_mutex> lock(mutex); boost::upgrade_to_unique_lock<boost::shared_mutex> write_lock(lock); } ... produces this error when compiled with Visual C++ 2010: C:\Users\braden>cl /I include/boost-1_43 /MD /EHsc upgrade_to_unique_lock_test.cpp Microsoft (R) C/C++ Optimizing Compiler Version 16.00.30319.01 for x64 Copyright (C) Microsoft Corporation. All rights reserved. upgrade_to_unique_lock_test.cpp include/boost-1_43\boost/thread/locks.hpp(823) : error C2679: binary '=' : no operator found which takes a right-hand operand of type 'boost::unique_lock<Mutex>' (or there is no acceptable conversion) with [ Mutex=boost::shared_mutex ] include/boost-1_43\boost/thread/locks.hpp(659): could be 'boost::upgrade_lock<Mutex> &boost::upgrade_lock<Mutex>::operator =(boost::upgrade_lock<Mutex>&)' with [ Mutex=boost::shared_mutex ] include/boost-1_43\boost/thread/locks.hpp(710): or 'boost::upgrade_lock<Mutex> &boost::upgrade_lock<Mutex>::operator =(boost::detail::thread_move_t<T>)' with [ Mutex=boost::shared_mutex, T=boost::upgrade_lock<boost::shared_mutex> ] include/boost-1_43\boost/thread/locks.hpp(717): or 'boost::upgrade_lock<Mutex> &boost::upgrade_lock<Mutex>::operator =(boost::detail::thread_move_t<T>)' with [ Mutex=boost::shared_mutex, T=boost::unique_lock<boost::shared_mutex> ] while trying to match the argument list '(boost::upgrade_lock<Mutex>, boost::unique_lock<Mutex>)' with [ Mutex=boost::shared_mutex ] include/boost-1_43\boost/thread/locks.hpp(820) : while compiling class template member function 'boost::upgrade_to_unique_lock<Mutex>::~upgrade_to_unique_lock(void)' with [ Mutex=boost::shared_mutex ] upgrade_to_unique_lock_test.cpp(7) : see reference to class template instantiation 'boost::upgrade_to_unique_lock<Mutex>' being compiled with [ Mutex=boost::shared_mutex ] I've searched a bit for previous reports of this sort of thing and I haven't found any. Hopefully that means I'm just doing something wrong. -- Braden McDaniel <braden(a)endoframe.com>

7 15

selective enabling/disabling of BOOST_AUTO_TEST_CASE
by Suresh Kumar 16 Sep '10

16 Sep '10

Hi I have several tests in my unit test file each of them uniquely named using the function BOOST_AUTO_TEST_CASE. I may not need to run all of them always and as of now what I do is to comment out the tests which I dont want to run. Instead of commenting out code, is there any other way I can selectively enable or disable tests if i have created tests with BOOST_AUTO_TEST_CASE. thanks suresh -- R Suresh Kumar, http://www.ee.ucr.edu/~sramachandranna -- Dhyaana–moolam Guror murtih Puja-moolam Guroh padam Mantra – moolam Guror vakyam Moksha – moolam Guroh krpa.

2 1

[Parameter] abstract base class
by Noah Roberts 15 Sep '10

15 Sep '10

I have a design in which I must inherit from a class with a whole lot of construction parameters. I need to have a pure virtual function in the class that inherits from this thing and is then provided by subclasses. Since the classes in question must all accept this huge list of construction parameters I would like to use named parameters. Unfortunately there's a problem caused by the attempt to create a function pointer declaration that takes the base class by value. I don't really understand why this is going on. Instead of just calling the base's constructor directly, the BOOST_PARAMETER_CONSTRUCTOR macro creates a type resolution instantiation with this: boost::parameter::aux::unaryfunptr_arg_type< void(*)(B) >::type 'B' being the base class. This explodes if B is an abstract class. Both g++ 3.4.5 (mingw) and VS2010 hate this line. Before I go about attempting to devise a workaround, what is the purpose of this strange bit of code? Why not just use 'B' directly?? The type resolution metafunction simply returns the 'B' place argument as its type anyway... I note some weird workaround code for ancient versions of MSVC but they seem to be created to make this type resolution thing work, not because 'B' is not known or something. However, I figure the people who wrote this library are smarter and more knowlegeable than I so the fact that I see no need for this code is probably my problem and not a reflection of it being unnecessary. -- http://crazyeddiecpp.blogspot.com/

2 2

Read from serial devices on USB
by Alejandro Martínez 15 Sep '10

15 Sep '10

Hi, I'm trying to communicate with a usb serial device plugged to a USB port. I've seen some examples with boost::asio and connecting to ie: /dev/ttyUSB0 for linux, or COM1 for windows etc. But i need something more robust and multiplatform. Like listing all the available devices and choosing by looking at the vendor id, or device id or something like that. How can i do that? Can this same approach be used later for say... bluetooth serial devices? How? Thank you all in advance. Alejandro

3 6

M68K-Coldfire: Boost 1.44 seems to have a problem with low level locking
by Andreas Wehrmann 15 Sep '10

15 Sep '10

Hello, I recently upgraded boost on my embedded linux platform (m68k coldfire architecture) from version 1.41 (cmake release) to 1.44. Boost itself compiles flawlessly but when I tried to link my application (which makes use of the thread, regex, system libraries) it fails with (note the last line): squ_loopback_timer.o: In function `boost::asio::detail::wait_handler<boost::_bi::bind_t<void, boost::_mfi::mf1<void, rapp::squ_loopback_timer_t, boost::system::error_code const&>, boost::_bi::list2<boost::_bi::value<rapp::squ_loopback_timer_t*>, boost::arg<1> > > >::do_complete(boost::asio::detail::task_io_service*, boost::asio::detail::task_io_service_operation*, boost::system::error_code, unsigned int)': squ_loopback_timer.cpp:(.text._ZN5boost4asio6detail12wait_handlerINS_3_bi6bind_tIvNS_4_mfi3mf1IvN4rapp20squ_loopback_timer_tERKNS_6system10error_codeEEENS3_5list2INS3_5valueIPS8_EENS_3argILi1EEEEEEEE11do_completeEPNS1_15task_io_serviceEPNS1_25task_io_service_operationESA_j[boost::asio::detail::wait_handler<boost::_bi::bind_t<void, boost::_mfi::mf1<void, rapp::squ_loopback_timer_t, boost::system::error_code const&>, boost::_bi::list2<boost::_bi::value<rapp::squ_loopback_timer_t*>, boost::arg<1> > > >::do_complete(boost::asio::detail::task_io_service*, boost::asio::detail::task_io_service_operation*, boost::system::error_code, unsigned int)]+0x72): undefined reference to `__sync_lock_test_and_set_4' What changed about the locking mechanism? Btw. I'm also using boost version 1.38 in another project (but on the same platform). My compiler is g++ version 4.2.0 20070318 (prerelease) (Sourcery G++ Lite 4.2-35). Any hints or help is appreciated! Best regards -- Dipl.-Ing. (FH) Andreas Wehrmann Software Development -------------------------------------------------------------- Center Communication Systems GmbH A-1210 Wien, Ignaz-Köck-Straße 19 Sitz in Wien FN 796 88p, Firmenbuchgericht Wien www.centersystems.com Tel.: +43 (0) 190 199 - 3616 Mobile: +43 (0) 664 884 75916 Fax: +43 (0) 190 199 - 2110 E-Mail: a.wehrmann(a)centersystems.com ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program.

3 5

[asio] io_service::work and cpu load
by Daniele Barzotti 15 Sep '10

15 Sep '10

Hi, in my DLL I create an endless io_service::work object to assign to a serial_port. I noted that, when I load the application that uses this DLL my CPU load raise up to 50% only when the io_service is created. Any suggestions? regards, Daniele.

1 0

Building PBGL
by David Doria 15 Sep '10

15 Sep '10

I downloaded PBGL from here: http://osl.iu.edu/research/pbgl/ <http://osl.iu.edu/research/pbgl/>When running CMake, it said that it couldn't find boost_mpi. I read that I need to do ./bjam --with-mpi I did that and it seemed to be successful (though very very quick): -------------------- Component configuration: ... - mpi : building ... ...found 1 target... The Boost C++ Libraries were successfully built! -------------------- So my questions are: 1) Does PBGL not come with the normal boost package from sourceforge ( http://sourceforge.net/projects/boost/files/boost/1.44.0/)? 2) How can I tell if boost_mpi built properly? Thanks, David

2 1

Re: [Boost-users] Why doesn't boost::function use get_pointer() ?
by Peter Soetens 15 Sep '10

15 Sep '10

On Tuesday 14 September 2010 23:47:41 Steven Watanabe wrote: > AMDG > > Peter Soetens wrote: > > When using boost::bind, I can write this: > > > > struct X > > { > > void foo(int) {} > > }; > > > > shared_ptr<X> x( new X ); > > // bind takes shared_ptr: > > bind(&X::foo, x, 3)(); > > > > But I can't write this: > > > > function<void(X*,int)> foo = &X::foo; > > // function refuses shared_ptr: > > foo(x,1); > > > > Why doesn't boost::function have the same get_pointer() trick ? > > Overload resolution for calling a boost::function is exactly the > same as for calling a function with that signature. shared_ptr<X> > is not convertible to X*, so the call fails. So this is by design I presume. > > > Or how can I > > rewrite the function signature such that it does work with shared > > pointers ? > > function<void(const shared_ptr<X>&, int)> foo; Thank you. The problem was that I didn't know if the first argument was a regular pointer or a shared pointer (generic code!). I have also found another solution, that is, by 'always' writing the call like this: foo(get_pointer(x),1); Which will work for all cases of x, and the signature of foo can remain void(X*,int). Peter

1 0