September 2011 - Boost-users

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

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

30 Sep '11

I have been using boost 1.45 on HP-UX using the acc compiler. Have used the following libs: asio, serialisation, file-system,data-time,program-options, shared_ptr spirit(classic) You need to test each library yourself. since as you have discovered boost is not officially supported on HP-UX. It appears as if HP-UX is no longer develop's/maintains their compilers.? Best regards, Ta, Avi

2 2

Boost.Test - how to turn off memory dump?
by John M. Dlugosz 30 Sep '11

30 Sep '11

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? —John

1 0

problem with xpressive sregex and stl::list::iterator
by Richard 30 Sep '11

30 Sep '11

hi, I am trying to use stl::list<boost::xpressive::sregex>, but the following code cannot be complied stl::list<boost::xpressive::sregex> rxs; sregex rex = boost::xpressive::sregex::compile("hello"); rxs.push_back(rex); 50 std::list<boost::xpressive::sregex>::iterator it; 51 52 for (it= rxs.begin(); it != rxs.end(); it++) 53 { //do something } line 52, "it= rxs.begin()" cannot be complied, it said the no match operator =., what is wrong? by the way, I cannot use std::vector<boost::xpressive::sregex>, it compiled failed. I searched on web, it is said there is a bug with this issue. I do not want to change the version of boost library I used, so let's ignore this issue at the moment. thanks. Richard.

3 3