- Boost - lists.preview.boost.org

[Asio] Match condition based on asio::streambuf length
by Chet S 27 Oct '08

27 Oct '08

Hello, I'm attempting to write a TCP-based server that can interpret a protocol that has a prepended length followed by variable length data. To do this, I'm using async_read_until along with an asio::streambuf. In order to ensure that there is enough data in the buffer for my next async read handler in the chain, I've written a match condition that will return true if there is enough data already in the streambuf to satisfy the read: class read_until_length { public: explicit read_until_length(size_t len) : m_len(len) { } template<typename Iterator> std::pair<Iterator, bool> operator()(Iterator begin, Iterator end) const { Iterator i = begin; const size_t buffer_len = std::distance(begin,end); if(buffer_len >= m_len) { std::advance(i, m_len); return std::make_pair(i, true); } else return std::make_pair(end, false); } private: size_t m_len; }; And then I use it in chained calls to async_read_until as such: Connection::start() { // Call our next handler when we have received at least 4 bytes (which represents the payload length that will follow) boost::async_read_until(m_socket, // boost::asio::ip::tcp::socket m_data, // boost::asio::streambuf read_until_length(4), boost::bind(&Connection::read_element, this, boost::asio::placeholders::error)); } Connection::read_length(boost::system::error_code& error) { // Read out the 4 byte size header size_t length = 0; memcpy(&length, &*boost::asio::buffers_begin(m_data.data()), sizeof(size_t)); m_data.consume(sizeof(size_t)); // Read out the variable length data next now that we know the size of it boost::asio::async_read_until(m_socket, m_data, read_until_length(length), boost::bind(&Connection::read_data this, length, boost::asio::placeholders::error)); } Connection::read_data(size_t length, boost::system::error_code& error) { // Consume the payload of length bytes... } The problem that I'm having is that async_read_until() performs two checks on the match condition. The first (boost/asio/impl/read_until.ipp:954) is to see if the iterator returned from the match condition is at the end of the buffer. If that is true, it does not post the next handler and assumes that not enough data has been read from the socket. This causes a problem when there is exactly enough data in the streambuf with the match condition as written above. It seems that returning true from the match condition (in the second part of the std::pair<>) should still post a handler since the condition was matched and it shouldn't be overridden by the iterator being at the end of the buffer. Is there a way to use async_read_until and a custom match condition to match the exact size of the streambuf? Thanks in advance. --Chet

1 0

Using allocator rebind on an incomplete type
by Chris Newbold 27 Oct '08

27 Oct '08

In trying to clean up some code in Boost.Pool, I've run afoul of some code in Microsoft's implementation of the STL where they are using allocator::rebind on an incomplete type. Something like this: struct Node; // Incomplete! template<typename T, typename Alloc> struct List { Alloc<T>::rebind<Node>::other::pointer node_p; ... } I believe that attempting to instantiate an allocator with an incomplete type is illegal so, by extension, I would have assumed that attempting to rebind on an incomplete type would also be illegal, no? -Chris

3 6

[threadpool] version 11 - some code cleanup
by k-oli＠gmx.de 27 Oct '08

27 Oct '08

Hello, I've uplaoded v11 to boost vault /concurrent programming. - documentation now in quickbook format, some enhancements - examples with Jam file - test with mJam file - code cleanup - API cleanup regards, Oliver

1 0

[test] linking does not return
by k-oli＠gmx.de 27 Oct '08

27 Oct '08

Hello, I've some test files but the link-step does not return (waits forever). gcc.link ../../../bin.v2/libs/tp/test/test_bounded_queue_fifo.test/gcc-4.3.1/debug/link-static/threading-multi/test_bounded_queue_fifo The Jam file: import testing ; project : requirements <library>../../test/build//boost_unit_test_framework <library>../../thread/build//boost_thread <library>../../system/build//boost_system <link>static <threading>multi ; rule tp-test ( source ) { return [ run $(source).cpp ] ; } test-suite thread_pool : [ tp-test test_bounded_queue_fifo ] [ tp-test test_bounded_queue_lifo ] [ tp-test test_bounded_queue_priority ] [ tp-test test_bounded_queue_smart ] [ tp-test test_unbounded_queue_fifo ] [ tp-test test_unbounded_queue_lifo ] [ tp-test test_unbounded_queue_priority ] [ tp-test test_unbounded_queue_smart ] ; Platform LINUX (Debian/Lenny). regards, Oliver

3 3

Preview 3 of the Geometry Library
by Barend Gehrels 27 Oct '08

27 Oct '08

Dear list, Herewith the URL to the third preview of our Geometry Library, including documentation and examples. The first two previews were sent to this list in January and in March. Since then, Bruno Lalande, already a Boost contributor, joined and we together reworked the library and its concepts. Lines (sequences of points) are now generic, handled using iterator pairs. Concepts have been worked out and integrated everywhere in the code. A Google Summer of Code project, handling Spatial Indexing, is based on the geometries of this library. Many discussions and comments were related to the concepts and, more specific, to the Point concept. This is the most basic thing of a geometry library and has to be as generic as possible. We think we implemented it like that into the library. The library can handle all point types including legacy point types, point types provided by the library, user defined point types, Cartesian points, native arrays, points with spherical coordinates. The library provides point_xy for Cartesian systems and point_ll for lat long systems, but the library user is thus not confined to those types. Using a strategy mechanism, the library should be able to handle every point type, and lines made of those types. The distance / simplify algorithms clearly show this design. The library is not completed, it is still a preview. It is used in real life projects but will still be enhanced and extended. The point type is now generic and lines (sequences of points) are generic. However, boxes, circles and polygons still have to be reviewed and their concepts can be worked out further. Some algorithms will be reprocessed to use strategies and handle any point type. The URL with documentation is here: http://geometrylibrary.geodan.nl. The library is Open Source and uses the Boost License. However, being a preview and not yet a real submission, the SVN is not at boost but still hosted at Geodan. If you are interested you can get SVN access. There were since March two other geometry library references sent to the list. This really shows the need for a Geometry Library into the boost collection. We welcome any discussion, comment or contribution and we are open for any type of cooperation, merge or join. Barend Gehrels

13 49

[zip_iterator] Understanding why.
by Robert Jones 27 Oct '08

27 Oct '08

Zip iterator implements the functionality I need to advance associated iterators through different ranges, but it seems strangly awkward to use, which is odd for Boost. so I'm wondering if I've misunderstood the design paradigms. Given two ranges (say vectors), I'd like to be able to generate an iterable composite, anditerate through that, something like vector<int> v1; vector<int> v2; typedef zip_composite<int, int> Z; Z zip( v1, v2 ); for ( Z::iterator i=zip.begin( ); i !=zip.end( ); ++i ) { // ... blah, blah, blah } Is this impossible or undesirable to implement for reasons I've not realized? Thanks, Rob.

4 3

[1.37.0] Release branch post-beta merges
by Beman Dawes 27 Oct '08

27 Oct '08

The release branch is is now open for post-beta merges, but only as described in the Post-Beta Merge Policy. See http://svn.boost.org/trac/boost/wiki/ReleasePractices/PostBetaMerges The final release is currently planned for Sunday, November 2nd. It would be helpful if all merges were complete a couple of days before that to allow plenty of time for tests to cycle and other pre-release preparation. Thanks, --The release managers

1 0

[release management] Draft policy for changes between beta and release
by Beman Dawes 27 Oct '08

27 Oct '08

In another post, John Maddock asked what sort of QA problems should be fixed between the beta and final releases, and what the criteria and procedures are. Here is a quick draft of a possible policy: * Showstoppers. Criteria: Bugs or other problems so serious that the release is seriously compromised if not fixed. Procedure: After discussion on the developers list, release managers will specify how the problem is to be attacked, and what effect it will have on schedule. * Documentation and other minor fixes not affecting code. Criteria: Changes not requiring regression testing and unlikely to impact other libraries. Procedure: Merges to branches/release are OK, after fix applied to trunk, and do not require a release manager's permission. * Minor code fixes. Criteria: Changes believed to have limited risk and that have already been made to trunk and are stable on trunk regression testing. Procedure: Merges to branches/release after fix stable on trunk tests do not require a release manager's permission. * Questionable code fixes. Criteria: Any code changes not meeting the "Minor code fix" criteria, such as a change with a lot of possible ripple effect on other libraries. Procedure: Ask release managers for permission. * Tools changes and fixes. Code changes to bjam, Boost.Build, Boost.Test, the doc build process, and other tools are discouraged late in release cycles in general and between beta and release in particular. Ask release managers for permission if you really think the change is warranted for this release. * New features, breaking changes, major reworks, new libraries. These are not suitable for merging between beta and release. Wait until the next release. Comments? --Beman

5 8

[1.37] status/Jamfile.v2
by Nicola Musatti 27 Oct '08

27 Oct '08

Hallo, I believe the Jamfile.v2 in the status directory should be merged to the release branch. The one that's there looks older than the directory reorganization that was performed a while ago, and it doesn't include the tests for several libraries. Cheers, Nicola -- Nicola.Musatti <at> gmail <dot> com Home: http://nicola.musatti.googlepages.com/home Blog: http://wthwdik.wordpress.com/

9 17

[mpl] Error in map docs (or bug)
by John Bytheway 26 Oct '08

26 Oct '08

I think I've found an error in the MPL docs (In trunk and 1.35, and presumably everything in between). The documentation for mpl::map state that "count<m,k>::type" yields "The number of elements with the key k in m", where in fact what it returns is the number of elements *equal* to k in m. This is consistent with the documentation of mpl::count. I suspect this is an error in the docs rather than a bug, since there is also mpl::has_key, which serves the purpose count claims to be serving (but isn't). Consider this code: #include <boost/mpl/has_key.hpp> #include <boost/mpl/count.hpp> #include <boost/mpl/assert.hpp> #include <boost/mpl/set.hpp> #include <boost/mpl/map.hpp> #include <boost/mpl/insert.hpp> using namespace boost::mpl; typedef set< int >::type my_set; typedef map< boost::mpl::pair<int, int> >::type my_map; BOOST_MPL_ASSERT((has_key<my_set, int >::type)); BOOST_MPL_ASSERT((count<my_set, int >::type)); BOOST_MPL_ASSERT((has_key<my_map, int >::type)); BOOST_MPL_ASSERT((count<my_map, int >::type)); BOOST_MPL_ASSERT((count<my_map, pair<int,int> >::type)); The fourth assertion fails. If the map docs were accurate then it would be the fifth assertion that failed. The same error is in the Associative Sequence docs, and perhaps other places. John Bytheway

1 0

Directory monitor for Windows (Boost.Asio extension)
by Boris 26 Oct '08

26 Oct '08

I had created a file monitor in March as a reference implementation of how to develop a Boost.Asio extension. While it was pretty incomplete at that time it has become a grown-up and fully tested directory monitor which supports both synchronous and asynchronous I/O. It currently works on Windows only as it is based on Microsoft's I/O Completion Ports (for those who need a directory monitor on Linux see http://article.gmane.org/gmane.comp.lib.boost.asio.user/2699) The directory monitor can be downloaded here: http://www.highscore.de/boost/dir_monitor.zip The zip archive contains various tests which can be used for a quick start. As the directory monitor is pretty small it shouldn't take much longer than 10 minutes to use it in your own projects. Boris

1 0

[test] Patch for gcc warning suppression.
by John Maddock 25 Oct '08

25 Oct '08

I'm getting a bunch of warnings from floating_point_comparison.hpp with gcc, the attached patch fixes these, if these are OK, I can commit from here. That just leaves: c:\data\boost\boost\trunk/boost/test/floating_point_comparison.hpp:246: warning: 'boost::test_tools::<unnamed>::check_is_small' defined but not used Is there any way to get rid of that as well? Thanks, John.

3 3

Welcome to the Boost Developers Archive on Google Groups!
by Eric Niebler 24 Oct '08

24 Oct '08

I've subscribed Google Groups to the boost developers list, so that people who would like to follow the group that way (and get an RSS feed to the list) can do so. http://groups.google.com/group/boost-developers-archive It can't import old messages, but new messages should show up there and be search-able. I hope. -- Eric Niebler BoostPro Computing http://www.boostpro.com

2 2

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

Re: [boost] [1.37.0][serialization] Assertion failed at libs/serialization/src/extended_type_info.cpp:92
by Hartmut Kaiser 24 Oct '08

24 Oct '08

Robert, Would it be possible to get an answer to my initial questions, please? Thanks! Regards Hartmut > Anything I can do about this? > What does this assertion mean? > > My setup: Linux 64Bit, gcc 4.2.4, serialization is used for polymorphic > objects loaded from different shared libraries, shared libraries are > loaded with dlopen(RTLD_GLOBAL|RTLD_LAZY). > > Assertion (extended_type_info.cpp:92) occurs during program termination > (after main returned), during dlclose(). > > If anything else isn't possible, would it be appropriate to change the > related code (before the release!) from: > > assert(start != end); > > // remove entry in map which corresponds to this type > do{ > if(this == *start) > x.erase(start++); > else > ++start; > }while(start != end); > > To: > > assert(start != end); > > // remove entry in map which corresponds to this type > while(start != end) { > if(this == *start) > x.erase(start++); > else > ++start; > } > > Which wouldn't remove the problem, but at least avoids hanging the > program if assertions are off... > > Regards Hartmut

3 6

[release management] Drop .gz and .zip release packaging?
by Beman Dawes 23 Oct '08

23 Oct '08

Boost releases are currently packaged four ways. With CR/NL line endings: boost_1_37_0_beta1.7z 27,340,204 boost_1_37_0_beta1.zip 57,017,457 With NL line endings: boost_1_37_0_beta1.tar.bz2 29,494,538 boost_1_37_0_beta1.tar.gz 36,436,716 I'd like to drop the .zip and .gz packages, starting with release 1.38.0. The primary motivation is a personal one; as release manager I do a lot of FTP transfers and would like to spend less time waiting for the big .zip and .gz packages to upload/download. The problem I see is that CR/NL users still download .zip files twice as often as .7z, and NL users still download about as many .gz packages as bz2. Comments? --Beman

11 11