- Boost-users - lists.boost.org

identity function
by Krzysztof Żelechowski 02 Oct '11

02 Oct '11

I use the following function on a regular basis: template < class P_T > P_T const &same (P_T const &p) { return p; } I use it for two purposes: 1. to pass an identity function to an API that expects a unary function, 2. to help the compiler identify the right chain of type conversions. The standard uses static_cast for (2), which could be used but it disables some compile-time verification so I would rather not use static_cast when not absolutely necessary. I have been unable to find a corresponding feature in Boost (although the value factory comes close). Do you think adding such a function to Boost utilities would be worthwhile? Chris

5 4

[boost-newbie] g++ Version 3 (MinGW) and compiling boost
by Robert Hartmann 01 Oct '11

01 Oct '11

Hi there, does anyone could say me how can I compile Boost.Filesystem with MinGW: Is something wrong with my b2-call see below, or is my project-config.jam buggy? ========================= import option ; using gcc ; using python : 2.7 # version : E:/Programme/Python27 # cmd-or-prefix : E:/Programme/Python27/include # includes : E:/Programme/Python27/libs # libraries ; using mpi : : <find-static-library>mpi <library-path>"E:/Programme/MPICH2/lib" <include>"E:/Programme/MPICH2/include" : "\"E:\\Programme\\MPICH2\\bin\\mpiexec\"" ; option.set keep-going : false ; ========================== g++ (GCC) 3.4.2 (mingw-special) Copyright (C) 2004 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. gcc (GCC) 3.4.2 (mingw-special) Copyright (C) 2004 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. I got as error message for Boost.Filesystem: ./boost/filesystem/v3/config.hpp:49:5: #error Configuration not supported: Boost.Filesystem V3 and later requires std::wstring support That means Boost.Filesystem V2 should work? Can I compile it without other boost things? I run set BUILDPATH=G:/Downloads/Libraries/boost_1_47_0/ownbuild set INSTALLPATH=G:/Downloads/Libraries/boost_1_47_0/distrib b2 --prefix=%INSTALLPATH%/libs/boost-1.47.0/distrib ^ --toolset=gcc ^ --build-dir=%BUILDPATH% ^ --layout=versioned ^ --variant=debug,release ^ --runtime-link=static ^ --link=shared,static ^ --threading=multi ^ --without-mpi ^ --without-python stage >build.log 2>&1 And in build.log I found many errors after waiting some Minutes: ...failed updating 26 targets... ...skipped 30 targets... ...updated 842 targets... If I doesn't use the options --without-mpi --without-python than about 30 seconds later I got this output in build.log: ====================== G:/Downloads/Libraries/boost_1_47_0/tools/build/v2/build\virtual-target.jam:1079: in virtual-target.register-actual-name from module virtual-target error: Duplicate name of actual target: <pstage\lib>mpi.pyd error: previous virtual target { common%common.copy-mpi.pyd.PYTHON_EXTENSION { gcc%gcc.link.dll-mpi.pyd.PYTHON_EXTENSION { gcc%gcc.compile.c++-python\collectives.o.OBJ { python/collectives.cpp.CPP } } { [...] error: created from ./stage-proper error: added properties: <debug-symbols>off <define>NDEBUG <inlining>full <library>object(file-target)@4418 <library>object(file-target)@4420 <library>object(file-target)@4488 <library>object(file-target)@4490 <library>object(file-target)@4602 <library>object(file-target)@4604 <library>object(file-target)@4617 <library>object(file-target)@4619 <library>object(searched-lib-target)@4507 <optimization>speed <runtime-debugging>off <variant>release <xdll-path>/G:/Downloads/Libraries/boost_1_47_0/ownbuild/boost/bin.v2/libs/mpi/build/gcc-mingw-3.4.2/release/threading-multi <xdll-path>/G:/Downloads/Libraries/boost_1_47_0/ownbuild/boost/bin.v2/libs/python/build/gcc-mingw-3.4.2/release/threading-multi <xdll-path>/G:/Downloads/Libraries/boost_1_47_0/ownbuild/boost/bin.v2/libs/serialization/build/gcc-mingw-3.4.2/release/threading-multi error: removed properties: <debug-symbols>on <inlining>off <library>object(file-target)@2472 <library>object(file-target)@2474 <library>object(file-target)@2542 <library>object(file-target)@2544 <library>object(file-target)@2656 <library>object(file-target)@2658 <library>object(file-target)@2671 <library>object(file-target)@2673 <library>object(searched-lib-target)@2561 <optimization>off <runtime-debugging>on <variant>debug <xdll-path>/G:/Downloads/Libraries/boost_1_47_0/ownbuild/boost/bin.v2/libs/mpi/build/gcc-mingw-3.4.2/debug/threading-multi <xdll-path>/G:/Downloads/Libraries/boost_1_47_0/ownbuild/boost/bin.v2/libs/python/build/gcc-mingw-3.4.2/debug/threading-multi <xdll-path>/G:/Downloads/Libraries/boost_1_47_0/ownbuild/boost/bin.v2/libs/serialization/build/gcc-mingw-3.4.2/debug/threading-multi G:/Downloads/Libraries/boost_1_47_0/tools/build/v2/build\virtual-target.jam:490: in actualize-no-scanner from module object(file-target)@5073 G:/Downloads/Libraries/boost_1_47_0/tools/build/v2/build\virtual-target.jam:135: in object(file-target)(a)5073.actualize from module object(file-target)@5073 G:/Downloads/Libraries/boost_1_47_0/tools/build/v2\build-system.jam:749: in load from module build-system G:\Downloads\Libraries\boost_1_47_0\tools\build\v2/kernel\modules.jam:283: in import from module modules G:\Downloads\Libraries\boost_1_47_0\tools\build\v2\kernel\bootstrap.jam:142: in boost-build from module G:\Downloads\Libraries\boost_1_47_0\boost-build.jam:17: in module scope from module ============================== Please, can any one help me? Best regards, Robert

2 2

question regarding Depth First Search algorithm
by Pablo Fleurquin 01 Oct '11

01 Oct '11

Hi, My name is Pablo Fleurquin.I am a student of M.Sc in Physics at the UIB (Palma de Mallorca, Spain) and I'm also working in the IFISC institute (Institute for Interdisciplinary Science and Complex Systems). I have a question regarding the Depth First Search algorithm. Ihave beenlooking at the Boost library page but I'm a newbie with this stuff. *Having a directed graph G (that could be a connected graph or not), I want my program to find the cycles within the graph and remove all the edges that are part of them. It doesn't matter which cycle is erased first as long as the result is the graph G without edge that formely were part of a cycle (in other words, the remaining edges are those that doesn´t form cycles) [attached is example1]* Intuitively, I am sure that using the depth_first_search() function and recalling it recursively for each "new" graph G "with less" cycles will finally arrive to graph G with no cycles. The problem is that I have the idea, but my knowledge of c++ is not so vast to implement this code. I would be gratefully, if you could show me a code for this implementation or a function that does that. Thank you. Pablo Fleurquin P.S: Example 2 is a situation that might happen. If the code can be implemented to eliminate these 2 cycles that share a link it would be greate, if not it will work fine as well.

3 13

class initialization problem
by Christopher Pisz 01 Oct '11

01 Oct '11

I have a Server class inherited from a Listener class, with the latter being bases on the boost::asio example code. My problem is that the Listener class requires am boost::asio::io_service & in order to initialize its member variables. I wanted the Server class to be responsible for starting a thread, owning the io_service, and running it. It seems I cannot do that! class Server : public Listener { public: Server() : Listener(ioService_) { } // Snip protected: boost::asio::io_service ioService_; }; class Listener { public: Listener(boost::asio::io_service & ioService) : socket_(ioService) { } // Snip protected: boot::asio::ip::tcp::socket socket_; }; The above throws an unhandled exception in the initialization of the derived class. I suppose this is because ioService_ is not constructed yet? Since the C++ rules, if I remember correctly, say the base class gets constructed before the members of the derived? So, I tried: Server() : ioService_(), Listener(ioService_) { } to try and force it, but it still throws an exception. The only way to get it to not throw an exception is to create the instance to io-service in main() and then pass it as a reference into the class constructor. If I do that, then there is no way for any class in my inheritance hierarchy to own the io_service. As such, I cannot start and stop it either. Is there any way to get ownership of the io_service in my class hierarchy?

3 2

[accumulators] Kahan algorithmic variants of sum_impl, mean_impl?
by Rhys Ulerich 30 Sep '11

30 Sep '11

>From a look at boost::accumulators::impl::sum_impl [1] it appears a straightforward, fast 'sum += value' is used to accumulate the running sum. That can fall prey to numerical errors if the sum is large and the value is small. A slower but more precise summation is possible using the Kahan summation algorithm [2]. Has anyone implemented a Kahan-based variant for either sum_impl or mean_impl [3]? Assuming not, other than to mimic existing code, any implementation suggestions? One gotcha for Kahan summation is that optimization levels employing additive associativity can spoil the computation. Are there other places in the Boost sources where this has become an issue and where I might look for a portable way around it? Thanks, Rhys [1] http://svn.boost.org/svn/boost/trunk/boost/accumulators/statistics/sum.hpp [2] http://en.wikipedia.org/wiki/Kahan_summation_algorithm [3] http://svn.boost.org/svn/boost/trunk/boost/accumulators/statistics/mean.hpp

2 1

Library for dealing with Unicode
by Jonathon 30 Sep '11

30 Sep '11

Hi all, I am looking for a library to help deal with locale in a "cross-platform" manner -- something like iconv but for cross-platform. After some searching, I've come across Boost.Locale, but it looks like it hasn't been released yet. Does anyone know when it will be accepted? Or, is there an alternative I can use? Thanks, J

3 2

Re: [Boost-users] Compilation of boost on HP-UX 11iv3 using aCC
by Avi Bahra 30 Sep '11

30 Sep '11

>> Which version of aCC are you using ? It must be lower version as compared to A.06.26. I searched on web and found above errors are not generated by its predecessor versions. You are right I am using: HP C/aC++ B3910B A.06.20 [May 13 2008] The newer compiler versions have too many regressions. If you cant use the version above, then you are out of luck. >>Have you tried gcc or acc ? My problem is that lower versions of aCC are licensed ( I am currently using trial version of aCC ) and to use latest version too much boost code >>needs to be hacked ! It is not at all a good idea. I have not tried later versions of gcc. What I have found is that old UNIX vendors(HPUX/IBM/ etc) are struggling with boost. This is not all surprising, since the boost developers do not have access. to these compilers, and even if they did, those compilers really struggle with advanced meta-programming techniques use in boost. Adding support for boost for them is a *major* undertaking, in time and resource. Hence they appear to have given up. From what I recall HPUX made a song and dance about support at boost 1.35, But have been silent ever since. If you are using any of these platforms, then do not make any assumption about boost. The organisation I work for is slowly migrating away from HPUX, in favour of linux. I suppose this is little consolation. Since you have a trial version of this compiler, can you submit the compilation failures as bugs/regressions ? Failing that you could isolate the boost libs you want to use, and see if you can fix the issues yourself or get help from the boost author. Either way it will likely entail some effort. Best regards, Ta, Avi

1 0

Re: [Boost-users] Boost.Test - how to turn off memory dump?
by Alex Perry 30 Sep '11

30 Sep '11

On Fri, 30 Sep 2011 00:05:49 -0500 John M. Dlugosz wrote :- > > If I don't care about global memory allocations being reported as "leaks", > how can I disable the "memory leak" detection step when a test program > completes? > --detect_memory_leak=0 As part of the command line arguments Alex

1 0

ipc message_queue getting permission denied
by Matt Fair 30 Sep '11

30 Sep '11

Hello, I ran across this problem when I moved my code to another system. Not sure if this is the right list to post this, but I've been having trouble debugging this as all of my attempts have lead me to dead ends. I have the following code with boost 1.47: try { m_messageQueue = boost::shared_ptr<boost::interprocess::message_queue>( new boost::interprocess::message_queue ( boost::interprocess::open_or_create ,name.c_str() //name ,numElements //max message number ,sizeof(Message) //max message size )); } catch(boost::interprocess::interprocess_exception &e) { cerr << e.what(); } where name="test_queue", numElements=100, and sizeof(Message)=256. The output is: Permission denied If I run as user root, it will pass. I can't figure out where it is trying to write to where it would have a permission error. Looking at the boost code, shouldn't it just be in /tmp/boost_interprocess which any user would have access to create and write to? Also, I noticedhttps://svn.boost.org/trac/boost/ticket/4250, but this should be fixed now. Any help or insight would be appreciated. Thanks, Matt

3 6

Status of boost::channel?
by niXman 30 Sep '11

30 Sep '11

All greetings! Tell to me please where I can learn about a library status boost::channel? Whether it will be accepted in the boost libraries? P.S. boost::channel: http://svn.boost.org/svn/boost/sandbox/channel/ Thanks. niXman.

1 0