January 2012 - Boost-users - lists.preview.boost.org

Re: [Boost-users] boost::asio::io_service::postand boost::asio::deadline_timer
by Marat Abrarov 18 Jan '12

18 Jan '12

Hi, Christopher. > I mean > now = current time > "Make biscuits 30 seconds from now." > "Clean the car 10 seconds from now." > "Ride your bike 45 seconds from now." > > Or does posting to the timer before it is "expired" interfere with the first > request? http://www.boost.org/doc/libs/1_48_0/doc/html/boost_asio/reference/basic_de… ~~~~~~~~~ Any pending asynchronous wait operations will be cancelled. The handler for each cancelled operation will be invoked with the boost::asio::error::operation_aborted error code. ~~~~~~~~~ Regards, Marat Abrarov.

2 1

[graph] Query for interest: implicit de Bruijn graph
by Jeremiah Willcock 18 Jan '12

18 Jan '12

Would any Boost.Graph users be interested if I implemented an implicit de Bruijn graph? The graph would have the following benefits and drawbacks: - Would not use any storage, but would act as a d^k-vertex, d^(k+1)-edge (for alphabet size d and word length k) graph. - Would be exactly the de Bruijn graph (with no vertex merging or filtering); those features could be added on top, using filtered_graph and/or another wrapper layer. - A filtered_graph based on a sparse property map could "enable" vertices and edges to process in a given algorithm. - A separate layer could enable processing of chains of vertices together. - I could provide a reverse complement map for the base-4 case, as well as conversions between vertices and edges and strings of the correct length. If anyone is interested, I have a few questions as well: - What should the performance tradeoff be between accessing a vertex's incident edges vs. computing reverse complements? - Does the k-mer size (word length) need to be chosen at run-time, or would a compile-time setting be acceptable? -- Jeremiah Willcock

1 0

Re: [Boost-users] [boost] Proposal for a Differential Evolution C++ library
by P Gregor 18 Jan '12

18 Jan '12

I could also help testing. regards, Piotr

1 0

how to calculate the Sample standard deviation using accumulators Hello,
by Victor Yankee 18 Jan '12

18 Jan '12

Hello, How can I calculate the Sample Standard Deviation over a std::vector of doubles using accumulators? Or is there a faster way in boost math or somesuch? Thanks

2 1

boost::asio::io_service::post and boost::asio::deadline_timer
by Christopher 18 Jan '12

18 Jan '12

I am getting boost:asio::error::operation_aborted error codes for an unknown reason in my handlers. No where did a cancel or reset occur, nor was the deadline_time or the io_service destroyed. Does anyone know if I may post to the same io_service a deadline_timer was constructed with or if they will interfere with eachother? May I post multiple times to the deadline timer calling back different methods? I can't imagine why the abort would show up other than I am taking something for granted.

2 2

compiler warnings in
by Chris Cleeland 17 Jan '12

17 Jan '12

Using Boost 1_47_0, in checking some automated builds on a CentOS 4 machine, I get the following errors from the compiler there (gcc v3.4.6). ../../../boost/exception/detail/error_info_impl.hpp:43: warning: `class boost::error_info<tag_answer, int>' has virtual functions but non-virtual destructor Indeed, if I go back and compile the boost tests for exceptions, I get the same compiler warning. GCC 3.4.6 is listed as a tested compiler for 1.47.0; is this warning then expected? Or not? The change that silences the compiler is declaring the destructor of boost::error_info_base to be virtual, but I'm not entirely sure if that's correct. I'm also hesitant to make this change only for CentOS 4 as that will change the object layout in memory and could result in linker confusion, though one could argue that linking against the wrong library is an error unto itself. If there's a more appropriate place to report/discuss these sorts of issues, please point me towards them. Thanks! -- Chris Cleeland

2 1

Jamfile.v2 for iostreams in Boost 1_48
by David Smith 17 Jan '12

17 Jan '12

Greetings, The Jamfile.v2 in the iostreams library of Boost 1_48 has the line (around 29): $(v) = [ modules.peek : $(v) ] ; Used to propogate path-constant settings to the local jam environment. I think it should be: $($(v)) = [ modules.peek : $(v) ] ; At least when I run with this the build of iostreams works correctly to compile bzip2 and zlib objects and fails otherwise. Has anyone else observed this? Regards David

3 2

boost::mpl, c++11 and clang 3.0: compiler bug?
by Allan Nielsen 17 Jan '12

17 Jan '12

Hi The following code compiles fine under clang 3.0 but I get unexpected output: #include <iostream> #include <boost/mpl/for_each.hpp> #include <boost/mpl/vector.hpp> template< typename _state > struct AccSize{ size_t s = 0; _state xxxx; template<typename T> void operator()(T) { std::cout << s << " " << sizeof(T) << std::endl; s += sizeof(T); }; }; struct Foo { Foo () {} Foo (const Foo & rhs) {} }; struct A{}; int main(int argc, const char *argv[]) { using namespace std; AccSize < Foo > p; typedef boost::mpl::vector<int, int, int> list; boost::mpl::for_each< list >( p ); return 0; } The output I get is: 0 4 0 4 0 4 The output I expected is: 0 4 4 4 8 4 If I remove the copy constructor in Foo, it works as I expect, and if I do a "normal" initialization of AccSize::s it also works as I expect. Is this a bug? If so, how can it be reproduced without mpl::foreach Best regards Allan W. Nielsen

3 3

[interprocess] hanging when open_or_create a mapped file
by yiling1 gong 17 Jan '12

17 Jan '12

hello, everyone: I’m a newbie to boost::interprocess, It’s really a cool library, so I try to write a simple program to test mapped file, it’s really simple: #include <boost/interprocess/managed_mapped_file.hpp> using namespace boost::interprocess; int main() { managed_mapped_file mfile(open_or_create, "/home/fractal/log.bin", 1024); //hanging here. return 0; } but it’s hanging on the creation of mapped file, I test it both on ubuntu 11.05(in virtualbox) and openSuse, the same problem remain. I’m sure the file exists, and I traced down the source code, it seems the program stuck in a while loop in function priv_open_or_create(...). except that, if I set the third argument of constructor from 1024 to 1, it’s even throw a exception( not good). any seggestion is appreciated!!!

1 0

boost::interprocess::detail::managed_open_or_create_impl::priv_open_or_create function
by Raghu Gyambavantha 16 Jan '12

16 Jan '12

Boost-ers, When I try to allocate more elements from a managed shared memory segment than allowed, I don't seem to be getting an error from the boost library. It looks like the reason for it is because the call to "construct_func" from "priv_open_or_create" in "boost::interprocess::detail::managed_open_or_create_impl" returns true/false, but that return value is never checked. I saw this error with boost versrion 1.43.0, but that specific portion of the code doesn't seem to have changed even in 1.48.0. >From the looks of it, it seems like the construct_func should throw instead of returning, but in 1.43.0, it is returning a value. Should the return value be check and appropriate action be taken here or am I missing something. Any help is much appreciated. Thanks, -Raghu

1 0