- Boost-users - lists.boost.org

thread/call_once compile error on trunk on my code
by Olaf Peter 22 Apr '13

22 Apr '13

Hello, my svn and compiler versions are: Path: . URL: http://svn.boost.org/svn/boost/trunk Repository Root: http://svn.boost.org/svn/boost Repository UUID: b8fc166d-592f-0410-95f2-cb63ce0dd405 Revision: 83985 Node Kind: directory Schedule: normal Last Changed Author: viboes Last Changed Rev: 83985 Last Changed Date: 2013-04-20 16:12:43 +0200 (Sat, 20 Apr 2013) $ LANG=en g++ --version g++ (GCC) 4.6.3 20120306 (Red Hat 4.6.3-2) Copyright (C) 2011 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. and maybe the last changes $ svn log boost/thread/pthread/once_atomic.hpp ------------------------------------------------------------------------ r83466 | viboes | 2013-03-16 16:36:21 +0100 (Sa, 16. Mär 2013) | 1 Zeile Thread: add invoke when no variadic templates are available; adapt call_once. breaks my code now which is basically: ---8<--- namespace { struct id_string { static boost::once_flag flag; typedef std::map<int, std::string> map_type; ... static void do_init(token_id_string& self) { insert(self.m_map)(x, "y")(...) ; } std::string const& operator()(id::type id) const { boost::call_once(flag, &id_string::do_init, *this); map_iterator s = m_map.find(id); ... } }; id_string::map_type id_string::m_map; #ifdef BOOST_THREAD_PROVIDES_ONCE_CXX11 boost::once_flag id_string::flag; #else boost::once_flag id_string::flag = BOOST_ONCE_INIT; #endif } --->8--- The error is about /usr/local/include/boost/thread/detail/invoke.hpp: In function 'typename boost::enable_if_c<(! boost::is_member_function_pointer<Fp>::value), Ret>::type boost::detail::invoke(Fp&&, Args&& ...) [with Ret = void, Fp = void (*)({anonymous}::id_string&), Args = {{anonymous}::id_string}, typename boost::enable_if_c<(! boost::is_member_function_pointer<Fp>::value), Ret>::type = void]': /usr/local/include/boost/thread/pthread/once_atomic.hpp:127:9: instantiated from 'void boost::call_once(boost::once_flag&, Function&&, ArgTypes&& ...) [with Function = void (*)({anonymous}::id_string&), ArgTypes = {{anonymous}::id_string&}]' /home/olaf/work/token_ids.cpp:93:59: instantiated from here /usr/local/include/boost/thread/detail/invoke.hpp:406:65: error: invalid initialization of non-const reference of type '{anonymous}::id_string&' from an rvalue of type '{anonymous}::id_string' Is this intentionally or do I miss something? Thanks, Olaf

2 6

[serialization] counting the xml cost
by Michael Powell 20 Apr '13

20 Apr '13

Hello, Basically, I'd like to find out how well the boost serialization facilities handle Xml and to what degree, robustness, etc. I am counting the cost of serialization for our model. We'll need do some basic CRUD type operations on our model objects throughout our system, and I want to use a nicely loosely coupled solution. Xml is the thought at first; I've also considered a lightweight database like SQLite which transfers easily for patches, upgrades, and other transfers. Looking at the boost serialization facilities, one keyword was emphasized in the tech-agnostic serialization documentation. Tech meaning whether this is a binary file, text, JSON, Xml, whatever. The keyword being that serialization and load occur in the same order. Enter the desire for Xml. One common problem in .NET C# land, at any rate, is that reads are commonly known as being asymmetric from writes. That is, not guaranteed to happen in the same order, composition, whatever. That is, you may see attributes in one instance, where as in another the element might have been broken out. Such as: <myObject name="something" /> and <myObject><name>something</name></myObject> are technically the same. The serializer might scoff at that, but I'm not sure the W3C suggest otherwise. That's my loose understanding of the specification. Regards, Michael Powell

1 3

Linker (/IDE) errors working with filesystem
by Michael Powell 20 Apr '13

20 Apr '13

Hello, I'm a novice (perhaps dangerously intermediate) boost user and have what I think is a linker error folding in filesystem, which would be a great thing to use for, well, file-system-y concerns in our software. However, am running into some strange conflicts between std::string and boost::filesystem. I can either use one or the other, but not both, without (what are possibly) "false positives" in the CodeBench IDE. We are targeting ArchLinux running on ARM cross compiled. As far as I know, I accepted the defaults building boost, excepting for python, etc. Linking against boost statically. I also ran across this article and wonder how I can ensure that we're linking in a correct order: http://www.boost.org/boost-build2/doc/html/bbv2/faq/liborder.html, if that's a plausible culprit / solution. I seem to recall there being auto-link support, but I'm not sure this will work in a toolchain like GCC (4.7.2). I want to say that possibly there is an IDE confusion using boost::filesystem with std::string, but is a tool chain issue not a boost issue. Regards, Michael Powell

1 1

[BGL] directed_graph_adapter
by Andreas Fabri 19 Apr '13

19 Apr '13

Hello, Did anybody write an adapter class (similar to filtered_graph) that gives a directed view on an undirected bidirectional graph. Or an adapter in the other direction, that is an undirected view on a directed graph. If not, is there a major obstacle, that is it simply cannot be implemented, even if it seems straightforward. best regards, andreas -- Andreas Fabri, PhD Chief Officer, GeometryFactory Editor, The CGAL Project phone: +33.492.954.912 skype: andreas.fabri

2 1

boost::cmatch indices
by Rob.Conde＠ai-solutions.com 19 Apr '13

19 Apr '13

Is there a reason why boost::cmatch.size() returns size_type, but boost::cmatch operator[] takes int? Thanks, Rob Conde

2 1

boost::asio::deadline_timer destruction
by Daniel 19 Apr '13

19 Apr '13

Hello, I've noticed that WaitHandler can be called after boost::asio::deadline_timer destruction. It may cause using of uninitialized data. Are there any other solutions of graceful destruction of boost::asio::deadline_timer to be sure that after this destruction WaitHandler will never be called? Example pseudo code: class UsingDeadlineTimer { char some_data_accessed_from_the_wait_handler[512]; boost::asio::deadline_timer timer; void WaitHandler(const boost::system::error_code &ec) { if(ec == boost::asio::error::operation_aborted) { return; } memset(some_data_accessed_from_the_wait_handler, 'E', 512); // cause the problem, because UsingDeadlineTimer may be already destroyed. } UsingDeadlineTimer() : timer(io_service) { timer.expires_at(the moment of UsingDeadlineTimer destruction); timer.async_wait( boost::bind(&UsingDeadlineTimer::WaitHandler, this, boost::asio::placeholders::error) ); } ~UsingDeadlineTimer() { timer.cancel(); // but we already can be in the WaitHandler body which will access some_data_accessed_from_the_wait_handler } }; Thank you, Daniel

2 5

mpl vector of pair access
by Olaf Peter 19 Apr '13

19 Apr '13

Hello, in these code I try to unpack the 1stand 2nd element, but it doesn't compile (here, it complains about »apply« in »struct boost::mpl::at_impl<boost::mpl::non_sequence_tag>« Is there a better way and how to get it to compile? Thanks, Olaf ---8<--- #include <boost/mpl/vector.hpp> #include <boost/mpl/vector_c.hpp> #include <boost/mpl/max_element.hpp> #include <boost/mpl/pair.hpp> #include <boost/mpl/integral_c.hpp> #include <boost/mpl/size_t.hpp> #include <boost/mpl/placeholders.hpp> #include <boost/mpl/at.hpp> #include <iostream> namespace mpl = boost::mpl; struct code { enum type { undefined, foo, bar }; }; typedef mpl::vector<std::size_t, // min max mpl::pair< mpl::size_t<0>, mpl::size_t<0> >, // undefined mpl::pair< mpl::size_t<1>, mpl::size_t<3> >, // foo mpl::pair< mpl::size_t<2>, mpl::size_t<5> > // bar > arity_info; template <code::type code> struct min_arity : mpl::at_c< typename mpl::at_c<arity_info, code>::type, 0 >::type { }; // same for max_arity at index 1 (pair::second) int main() { std::cout << "min arity = " << min_arity<code::bar>::value << '\n'; // std::cout << "max arity = " // << max_arity<code::bar>::value << '\n'; } --->8---

2 7

Recommended linker order
by Michael Powell 19 Apr '13

19 Apr '13

Hello, So I am reading the article: http://www.boost.org/boost-build2/doc/html/bbv2/faq/liborder.html But I am not clear what this means? For instance, I'd like to put filesystem to use, but was finding linker errors there. Either could use filesystem issues, or std::string issues (<string>), but not both, for instance. No clue how that could even be possible. Build was also failing along similar lines, semantic and/or what I gather are linker failures. So... I get to specify the order CodeBench for Linux targeting ARM, wants to link libraries. Is there a specific order to include boost static libs that keeps the system happy? If so, what? Thank you... Regards, Michael Powell

2 1

heap handle invalidation
by ☂Josh Chia (谢任中) 19 Apr '13

19 Apr '13

Hi, I'm using a mutable heap, in particular a mutable d_ary_heap, for the priority queue in Djikstra's algorithm. When does a handle become invalid? I don't know how a d_ary_heap is implemented, but if it's like the standard array-backed binary heap, then I would expect invalidation of many handles whenever an item gets popped or an item gets increased or decreased. I need references to the elements in order to update them, so if the invalidation I expect does happen as I describe, how can I get permanent references/pointers/iterators to the elements? Perhaps another implementation besides d_ary_heap? Although I haven't profiled the performance, I think even though something like the fibonacci heap may have permanent handles, pointer chasing, more frequent cache misses and frequent dynamic memory allocation may be detrimental to performance. Which heap implementation supports key increase without handle invalidation? Josh

1 0

boost.log: update filters at runtime
by Jonathan Walker 18 Apr '13

18 Apr '13

Hello, I've been working on integrating the boost logger into my application for the past few days now. I'm looking for the ability to update filters at runtime based on settings from a file or other source. A specific use-case would be the ability to decrease a severity level during a debugging session for more detailed log reports without having to restart the application. Is there anything like this available in the current release? Is there any plan to incorporate this kind of functionality in future releases? Thanks, Jonathan ________________________________ This e-mail and any attachments are confidential. If it is not intended for you, please notify the sender, and please erase and ignore the contents.

1 0