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

boost::date_time with BOOST_DATE_TIME_POSIX_TIME_STD_CONFIG on MinGW: segfault on multithreading
by Yana A Kireyonok (aka Iron Bug) 13 Jan '12

13 Jan '12

Hello, everybody. Looks like I found a bug in date_time library when building with MinGW. I've built static boost 1.48.0 libraries under windows (Win7 64) with MinGW 3.20 (32 bit, GCC 4.6.1) with BOOST_DATE_TIME_POSIX_TIME_STD_CONFIG option and threading=multi link=static runtime-link=static address-model=32 In this case, when using threads (-mthreads) simple operations with timings fail. I simply created a thread that slept for some milliseconds ten times next and it fails. I traced a little, it fails inside date_time functions, called from this_thread::sleep() with SIGSEGV, segmentation fault. That happens in ctor of structure simple_time_rep, out of reasons, at first glance. Looks like this structure or maybe standard time functions of MinGW are not reentrant (this is my guess, I did not dig too deep in the causes of segfault). Without flag BOOST_DATE_TIME_POSIX_TIME_STD_CONFIG it's going all right with timings and threads. But when using precise timings and threading it fails. I also tried flags BOOST_HAS_WINTHREADS and _REENTRANT - that does not affect the case. BOOST_THREAD_USE_LIB was defined (to let it work with static libs). Is this a bug or problem of MinGW of certain version? I haven't found anything similar in bugtracker. Sincerely, Yana A. Kireyonok

1 0

const char* in shared memory object
by Sabine B 13 Jan '12

13 Jan '12

Hi all! I'm a newbie with boost and I'm struggling with objects in shared memory. The following struct is given: struct MyType{ int id; const char* name; const char* rep; MyType(){}; MyType(MyType* const& type){ id = type->id; name = type->name; rep = type->rep; }; Objects of MyType should be shared between processes in shared memory. I have no problem to put and extract the objects from shared space. But I'm struggling with the const char* pointers. I know that pointers are not allowed in shared memory, thats why I tried to realize with boost string by doing the following: typedef boost::interprocess::allocator<char,boost::interprocess::managed_shared_memory::segment_manager> CharAllocator; typedef boost::interprocess::basic_string<char, std::char_traits<char>, CharAllocator> MyString; boost::interprocess::managed_shared_memory segment(boost::interprocess::open_only, "MySharedMemory"); CharAllocator charallocator(segment.get_segment_manager()); struct MyType{ int id; MyString name; MyString rep; MyType(){}; MyType(MyType* const& type){ id = type->id; name = type->name; rep = type->rep; }; But this is not working because the segment and charallocator are not recognized in the struct and if I put boost::interprocess::managed_shared_memory segment(boost::interprocess::open_only, "MySharedMemory"); CharAllocator charallocator(segment.get_segment_manager()); In the struct I get compiler errors. Could anybody help me how I can realize const char* in an object that should be shared? Please help me! Thank you very much! Best regards, Sabine

2 2

Boost.Move template-function parameter
by Luke J. West 12 Jan '12

12 Jan '12

Hi. I'm trying to implement a template function that 'moves' it's parameter. Here's a simple moveable class... struct my_class { my_class() {} my_class(BOOST_RV_REF(my_class)) {} my_class& operator=(BOOST_COPY_ASSIGN_REF(my_class) c) {return *this;} BOOST_COPYABLE_AND_MOVABLE(my_class); }; And here is the function... template<typename T > void steal(BOOST_COPY_ASSIGN_REF(T) c) {} so it works fine when I do this... int main() { my_class c; steal<my_class >(c); } but not when I do this... int main() { int c; steal<int >(c); } It's as if 'int' is not a movable type which of course it trivially is. I read that Boost.Container classes are 'move aware' and I assume boost::vector<int > works, so what am I doing wrong - how to write my 'steal' function capable of moving it's argument when it's moveable class, but not breaking for primitives? Thanks, Luke (hopeful newbie!)

2 1

Question regarding boost::program_options
by Ankur Sinha 12 Jan '12

12 Jan '12

Hello, I've just started using boost. I'm making use of the program_options to take arguments for my tool. I've written the following: > mDesc.add_options() > ("help,h", "Print this usage message.") > ... > .... > .. > ("v-level,v",boost::program_options::value<int>(&mVerboseLevel),"Debug level: 1,2,3\n(default: 0)") > ("version",boost::program_options::value<std::string>()->zero_tokens(),"Package information: version etc.") > and then, I have a method, which I call in main(int argc, char **argv): > /* > *-------------------------------------------------------------------------------------- > * Class: Session > * Method: Session :: ParseCommandLine > * Description: Parse the command line > *-------------------------------------------------------------------------------------- > */ > int > Session::ParseCommandLine (int argc, char **argv) > { > boost::program_options::store(boost::program_options::parse_command_line(argc, argv, mDesc), mVariableMap); > /* notify required to save the values into my variables. */ > /* TODO: confirm this behaviour */ > boost::program_options::notify(mVariableMap); > > if (mVariableMap.count("help")) > { > std::cout << mDesc << std::endl; > return -1; > } > return 0; > } /* ----- end of method Session::ParseCommandLine ----- */ > However, when I invoke my tool, I get an exception (which I admit I haven't handled yet): > [ankur@ankur dump]$ ./src/test --version > terminate called after throwing an instance of 'boost::exception_detail::clone_impl<boost::exception_detail::error_info_injector<boost::program_options::validation_error> >' > what(): in option 'version': at least one value required > Aborted (core dumped) > [ankur@ankur dump]$ > I'm referring this document[1] which says that zero_tokens() implies an argument may or may not be provided. Why is the exception coming up saying I need "at least one value"? What am I missing? Another query: is notify() necessary to save the values provided on the command line to the variables I pass in the add_options() from the variable map (to mVerboseLevel for instance)? [1] http://www.boost.org/doc/libs/1_42_0/doc/html/program_options/overview.html -- Thanks, Regards, Ankur: "FranciscoD" http://fedoraproject.org/wiki/User:Ankursinha http://dodoincfedora.wordpress.com/

1 0

[BGL] creating a single vertex for the loop in a graph
by srinivas boppu 12 Jan '12

12 Jan '12

Hello All, I am looking for the following in boost graph. I have a graph in which I am able to find out the loops. Now, I wanted to replace the whole loop with a single vertex in the same Graph. Is that possible ? if it is, can you suggest me how do it. Alternate way could be, I copy the loop and make another graph and add this whole graph as a "vertex" in initial graph. Can I add a "graph" as a vertex to existing graph? Thanks and regards, Srinivas Boppu

3 2

Boost, git or svn?
by Robert Jones 12 Jan '12

12 Jan '12

I'm confused about the current status of the svn -> git transition,if indeed any such transition is occuring. Boost main page still seems refer to the old SVN repository, but vault references are to a git-hub server. Can someone supply a few sentences of explanation. Thx. - Rob.

3 3

[Range] Range adaptors and compatibility with non-adapted iterators
by John M. Dlugosz 12 Jan '12

12 Jan '12

Consider something like this: themap_type::iterator search_for (int val) { return boost::find_if (themap|map_values, my_predicate(val)); } The return from the function chokes because the adapted iterator type cannot be convered back to the original type of iterator in the parent collection. This is more general. If I wasn't just returning through but using the result within the function, how do I test against end to see if find_if was successful? auto it2= boost::find_if (themap|map_values, my_predicate(val)); if (it2 == themap::end()) // same problem! This seems to be a problem with any algorithm that returns an iterator or otherwise makes the iterators used within the algorithm visible outside of it, even if such usage is just as an "end" flag and not really used to locate the element within the original collection for subsequent iteration. Am I missing something? Is there a good reason why the filtered iterator can't give up the inner original iterator? How do you use range adaptors in these situations? —John

4 6

Help needed in creating Regex...
by Subhash Nagre 12 Jan '12

12 Jan '12

Hi everyone, I need a help in creating a regular expression, I need to select only "Sentences" in String using Regex so I wrote the Code .. *boost::regex re( "(\\S.+?[.!?])(?=\\s+|$)" );* This code considers that sentences break at any of these three characters . ! ? but this code considers string "Mr. Obama" as 2 difference sentences (being a dot character in between), where as I want it to be only one sentence. I have a got a string vector of all the words which have to excluded. std::vector<std:string>exceptionVector; exceptionVector.pushback("Mr.") exceptionVector.pushback("Dr.") exceptionVector.pushback("Mrs.") Now in my regex, I want that if any of the words occur from my vector, it should NOT be a seperate sentence. Any idea how can I create this Regular expression Thanks in Advance -Subhash

3 2

asio raw with PF_PACKET and promiscuous mode?
by Chris Cleeland 11 Jan '12

11 Jan '12

Before I reinvent the wheel, has anybody put together an asio encapsulation of Linux' AF_PACKET sockets with SOCK_RAW? Also, I'm trying to wrap my brain around how I might represent the concept of an "endpoint" for a socket like this. Rather than an IPv4/v6 address, one binds to an interface. So, while there is a concept of a resolver, it resolves an interface name to an interface index value. >From a C/system call perspective, I need to do the following to get the ball rolling: 1. s = socket(AF_PACKET, SOCK_RAW, htons(ETH_P_ALL)) 2. sockaddr_ll myinterface = { .sll_family = AF_PACKET, .sll_protocol = htons(ETH_P_ALL), .sll_ifindex = interfaceIndex("eth0") }, then bind(s, &myinterface) 3. packet_mreq promiscuous_cmd = { .mr_ifindex = interfaceIndex("eth0"), .mr_type = PACKET_MR_PROMISC }, then setsockopt(s, PACKET_ADD_MEMBERSHIP, &promiscuous_cmd) I'm pretty sure that for #1, I need a class packet that models the Protocol concept. But where to put the rest is a bit fuzzier. Do I need to create a whole new "resolver" for interface names? I guess I should also note that, at least for the moment, this is just a means to an end, rather than the end itself. So, I don't need to craft the entire thing--I need to get enough working to be able to easily write some code that can capture all network packets on a given interface, e.g., class snooper { public: snooper(std::string interfacename); void start(); void stop(); ... }; so I can just do something like snooper s("eth0"); s.start(); ...capture stuff... s.stop(); (I'm obviously omitting the part where snooper would hook into objects that would receive and process captured packets, but I'd like for that to work similar to how one plumbs up receipt of udp packets.) -- Chris Cleeland

1 0

casting shared pointers to derived classes
by Don Wilde 11 Jan '12

11 Jan '12

Greetings, all - I am updating a neural network package from the DOS days to modern C++ standards. I have a shared_ptr to a vector of shared_ptrs of type Layer. This class has derived classes OutputLayer, MiddleLayer and InputLayer. Each of these has their own method for calculating output updates, so I need to be able to cast. from boost::shared_ptr<Layer> to static_cast<OutputLayer*>(). Unfortunately, VC++ 2010 will not allow me to do so. It won't even allow a reinterpret_cast. I've looked on the Web and I also have Schalling's guide to Boost. Thank you in advance for your guidance. I'm using boost_1_47_0, if that makes a difference. :D

4 4