October 2008 - Boost - lists.boost.org

Class-level instrospection
by Joel Falcou 24 Oct '08

24 Oct '08

Daniel Walker a écrit : > Why don't we both package what we have, upload to vault, and then > work out merging the two and adding the rest in another thread? I've cleaned-up my files and 'boostified' them so to speak. I've uploaded a 'introspection.zip' on the vault so you can grab it and we can start discussing this in a proper topic ;) Basically i've uploaded the following : BOOST_HAVE_MEMBER_FUNCTION BOOST_HAVE_CONST_MEMBER_FUNCTION BOOST_HAVE_NON_CONST_MEMBER_FUNCTION BOOST_HAVE_MEMBER_DATA Basically, the usage is really simple. You want to check if classes provides a bool is_ok(long) method and you want to catch both const and non-const variations : BOOST_HAVE_MEMBER_FUNCTION(is_ok, bool(long) ) then the check itself is done via : has_member_function<T>::type For BOOST_HAVE_DATA_MEMBER, my method can only check for public members. So I added a macoer called BOOST_INSTROSPECTION_SUPPORT(C,V) that have to be used in user-defined class C that want to publish a member named V to our introspection mecanism. struct foo { private : int mValue; }; struct bar { BOOST_INTROSPECTION_SUPPORT(bar,mValue); private : int mValue; }; BOOST_HAVE_MEMBER_DATA(int, mValue) has_member_data_mValue<foo>::value is false has_member_data_mValue<bar>::value is true Other limitations is that currently you can't have various check for a method with a given name but different signature. I'm working on a fix using PP sequence and : BOOST_HAS_MEMBER_FUNCTION_OVERLOADS(func,(void())(void(int))(void(int,int))) In the same way, checking for suff like operator[] yields invalid class name has_member_function_operator[]. So I propose we hand-code the following : BOOST_HAS_SUBSCRIPT_OPERATOR( void(int) ) that builds a has_subscript_operator<T> class. Other similar operator should be handled the same way.

4 7

[General] A standard idiom or technique?
by Robert Jones 24 Oct '08

24 Oct '08

A recent post in this list started me thinking... Given this code snippet, using namespace std; void f( int, int ); vector<int> v1; vector<int> v2; // populate... for ( unsigned i = 0; i <= v1.size( ); ++ i ) { f( v1[ i ], v2[ i ] ); } How would you write this as a for_each() or in some other Boosty, functional way? (The aspect I'm trying to highlight is that two iterators over different ranges must advance in step) One could imagine a tuple-based solution, in which one specifies how an operation applied to the tuple distributes to operations applied to each of the tuple's elements. Does any such thing exist in Boost? Is it ever necessary, or can the functionality be achieved by some C++/STL/Boost idiom that I've not thought of? If there is no obvious technique, is there any interest in a Boost library which address this in a generic way? Thanks. - Rob.

5 7

How to install Boost.Numeric.Bindings
by Peng Yu 24 Oct '08

24 Oct '08

Hi, I've downloaded the whole boost src. svn co https://svn.boost.org/svn/boost boost Right now, I need Boost.Numeric.Bindings which is in the sandbox. I tried bjam, but I got the following error. Can somebody let me know how to compile Boost.Numeric.Bindings? Thanks, Peng $ bjam Unable to load Boost.Build: could not find build system. --------------------------------------------------------- /home/local/eda06/yupeng/test/C++/svn/boost/sandbox/boost-build.jam attempted to load the build system by invoking 'boost-build ;' but we were unable to find "bootstrap.jam" in the specified directory or in BOOST_BUILD_PATH (searching /usr/share/boost-build). Please consult the documentation at 'http://www.boost.org'.

3 8

[Pool]: How to tag pool_allocator
by VoidEx 24 Oct '08

24 Oct '08

Hello pool_allocator based on singleton allocator, but uses it's own tag. But I want use one pool_allocator for some strings/vectors/lists and another for some other data, because I want destroy data separately. Is it possible now? Or may be I do it wrong?

2 2

[pool] Conversion to Boost.Thread for singleton_pool
by Chris Newbold 24 Oct '08

24 Oct '08

Boost.Pool contains a private mutex implementation used as the default mutex type for singleton_pool (which provides a thread-safe API). Now that Boost.Thread is well-established and in line with TR1, I'd like to jettison the private mutex implementation inside Boost.Pool and switch the default mutex type to be boost::mutex. The obvious consequence of this change is that singleton_pool becomes dependent on Boost.Thread. The dependency would exist only for singleton_pool and its clients and would not affect the other Boost.Pool interfaces. Anyone have any objections or reasons why this wouldn't be a good idea? -Chris

3 5

[Generator Iterator] [random]
by Andy Tompkins 24 Oct '08

24 Oct '08

I am trying to use boost::generator_iterator with a random number generator as follows: --- #include <boost/generator_iterator.hpp> #include <boost/random/mersenne_twister.hpp> class my_generator {...}; my_generator gen; boost::generator_iterator<my_generator> first(&gen); boost::generator_iterator<my_generator> last; boost::random::mt19937 rng; rng.seed(first, last); --- Sadly the default constructor of generator_iterator leaves m_gen uninitialized and the seed function is trying to compare first with last. I could pass gen to last, ie. boost::generator_iterator<my_generator> last(&gen), but then if ever begin has the same value as end, it will stop. The documentation for generator_iterator says that it models an input iterator, and I assume that an input iterator is ok to use with the seed function. I'm not sure what can/should be done. Thanks in advance, Andy.

2 3

[filesystem] file monitoring
by Carlos Rafael Giani 24 Oct '08

24 Oct '08

Hi, is a file monitor feature planned for boost.filesystem? I have been looking for a platform independent solution, but there is none (except in toolkits like Qt). Also, so far only inotify (a subsystem in the Linux kernel) can tell me when a file is closed after writing, which is absolutely necessary when watching a plugins/ directory, for example (you don't want to touch the plugins when they're only halfway copied). Several tools watch the plugin directory, but don't care if the plugins are being written or are fully there already. This can lead to instabilities. I think it is only logical to add this to boost.filesystem, something like set_monitor_callback(pathname, function_object); any thoughts?

10 11

[function] function.hpp lacks include guard
by Okko Willeboordse 24 Oct '08

24 Oct '08

Hello, My co-worker Bob Praster found, using PC-Lint 9.0, that function.hpp doesn't have an include guard like; #ifndef BOOST_FUNCTION_HPP #define BOOST_FUNCTION_HPP ... #endif // BOOST_FUNCTION_HPP Kind regards, Okko Willeboordse

12 17

[boost.org] people not updated?
by Thorsten Ottosen 24 Oct '08

24 Oct '08

Hi, It seems like we are missing some of our recent members. -Thorsten

4 3

[asio] HTTPS async_read() question
by Gregory Dai 24 Oct '08

24 Oct '08

I'm playing with the following SSL client example against an https site: http://www.boost.org/doc/libs/1_36_0/doc/html/boost_asio/example/ssl/client… I replaced std::cout << "Enter message: " with the part of the URL corresponding to an HTTP GET request, and changed the handle_read() to void handle_read(const boost::system::error_code& error, size_t bytes_transferred) { if (!error) { std::cout << "Reply: "; std::cout.write(reply_, bytes_transferred); std::cout << "\n"; boost::asio::async_read(socket_, boost::asio::buffer(reply_, bytes_transferred), boost::bind(&client::handle_read, this, boost::asio::placeholders::error, boost::asio::placeholders::bytes_transferred)); } else { std::cout << "Read failed: " << error << "\n"; } } so that it goes back to read more data. I'm able to get more than 20k data in response, but unable to reach the end of the data stream as the browser can. What could be a good reason for this behavior? Thanks, Greg

2 2