Search Results

Search found 1726 results on 70 pages for 'boost propertytree'.

Page 14/70 | < Previous Page | 10 11 12 13 14 15 16 17 18 19 20 21  | Next Page >

  • boost::enable_if class template method

    - by aaa
    I got class with template methods that looks at this: struct undefined {}; template<typename T> struct is_undefined : mpl::false_ {}; template<> struct is_undefined<undefined> : mpl::true_ {}; template<class C> struct foo { template<class F, class V> typename boost::disable_if<is_undefined<C> >::type apply(const F &f, const V &variables) { } template<class F, class V> typename boost::enable_if<is_undefined<C> >::type apply(const F &f, const V &variables) { } }; apparently, both templates are instantiated, resulting in compile time error. is instantiation of template methods different from instantiation of free functions? I have fixed this differently, but I would like to know what is up. Thank you

    Read the article

  • How to use the boost lexical_cast library for just for checking input

    - by Inverse
    I use the boost lexical_cast library for parsing text data into numeric values quite often. In several situations however, I only need to check if values are numeric; I don't actually need or use the conversion. So, I was thinking about writing a simple function to test if a string is a double: template<typename T> bool is_double(const T& s) { try { boost::lexical_cast<double>(s); return true; } catch (...) { return false; } } My question is, are there any optimizing compilers that would drop out the lexical_cast here since I never actually use the value? Is there a better technique to use the lexical_cast library to perform input checking?

    Read the article

  • Boost: Although compiled, missing lib

    - by Jonathan
    I've compiled Boost and am using it in a Visual C++ project. At some point, I started getting the following error: 1>LINK : fatal error LNK1104: cannot open file 'libboost_filesystem-vc100-mt-sgd-1_45.lib' AFAIK I have the linker configuration set correctly. I've searched for the lib and the closest I found was: libboost_filesystem-vc100-mt-gd-1_45.lib (Notice the missing 's') What does the missing 's' represent? What do these similar yet different libs stand for? How come I'm missing it although I've compiled all of Boost?

    Read the article

  • boost variant static_visitor problem picking correct function

    - by Steve
    I'm sure I'm having a problem with template resolution here, but I'm not sure why I'm having the problem. I have a static visitor I'm passing to boost variant where i've had to do template specialization for certain cases. The case for everything except for MyClass should throw in the static_visitor below. Unfortunately, when the visitor is applied to pull a MyClass out, it selects the most generic case rather than the exact match. I would type each case explicitly, but that will be rather long. So, why is the compiler resolving the most generic case over the exact match, and is there anyway to fix it template<> class CastVisitor<MyClass>:public boost::static_visitor<MyClass> { public: template<typename U> MyClass operator()(const U & i) const { throw std::exception("Unable to cast"); } MyClass operator()(const MyClass& i) { return i; } };

    Read the article

  • Boost.Log - Multiple processes to one log file?

    - by Kevin
    Reading through the doc for Boost.Log, it explains how to "fan out" into multiple files/sinks pretty well from one application, and how to get multiple threads working together to log to one place, but is there any documentation on how to get multiple processes logging to a single log file? What I imagine is that every process would log to its own "private" log file, but in addition, any messages above a certain severity would also go to a "common" log file. Is this possible with Boost.Log? Is there some configuration of the sinks that makes this easy? I understand that I will likely have the same "timestamp out of order" problem described in the FAQ here, but that's OK, as long as the timestamps are correct I can work with that. This is all on one machine, so no remote filesystem problems either.

    Read the article

  • boost test case for function taking user input

    - by oadams
    I have a function that takes in user input via std::cin: std::getline(std::cin, in); and creates a corresponding data structure by matching it with a regular expression. The function then returns this data structure. I'm using boost.test and I want to create a unit test to check that the output data type is correct given some inputs. However I don't know how to go about it since the input isn't passed as an argument to the function. EDIT: Is there a simple way to create a boost test case that feeds the function a string via standard input?

    Read the article

  • Linker Error : Statically Linking of Boost Serialization Library

    - by Manikanda raj S
    I'm trying to link the Boost Serialization Library to my Code. But it doesn't seem to be working. g++ serialize.cpp -L"/usr/local/lib/libboost_serialization.a" Error : /tmp/ccw7eX4A.o: In function boost::archive::text_oarchive::text_oarchive(std::basic_ostream<char, std::char_traits<char> >&, unsigned int)': serializep.cpp:(.text._ZN5boost7archive13text_oarchiveC2ERSoj[_ZN5boost7archive13text_oarchiveC5ERSoj]+0x25): undefined reference toboost::archive::text_oarchive_impl::text_oarchive_impl(std::basic_ostream &, unsigned int)' .......... collect2: ld returned 1 exit status But when i link as a shared library, g++ serialize.cpp -lboost_serialization , it works fine. What am i missing here P.S : Other StackOverflow posts with the same question has no answers that work for the above error

    Read the article

  • Can I use boost::make_shared with a private constructor?

    - by Billy ONeal
    Consider the following: class DirectoryIterator; namespace detail { class FileDataProxy; class DirectoryIteratorImpl { friend class DirectoryIterator; friend class FileDataProxy; WIN32_FIND_DATAW currentData; HANDLE hFind; std::wstring root; DirectoryIteratorImpl(); explicit DirectoryIteratorImpl(const std::wstring& pathSpec); void increment(); public: ~DirectoryIteratorImpl() {}; }; class FileDataProxy //Serves as a proxy to the WIN32_FIND_DATA struture inside the iterator. { friend class DirectoryIterator; boost::shared_ptr<DirectoryIteratorImpl> iteratorSource; FileDataProxy(boost::shared_ptr<DirectoryIteratorImpl> parent) : iteratorSource(parent) {}; public: std::wstring GetFolderPath() const { return iteratorSource->root; } }; } class DirectoryIterator : public boost::iterator_facade<DirectoryIterator, detail::FileDataProxy, std::input_iterator_tag> { friend class boost::iterator_core_access; boost::shared_ptr<detail::DirectoryIteratorImpl> impl; void increment() { impl->increment(); }; detail::FileDataProxy dereference() const { return detail::FileDataProxy(impl); }; public: DirectoryIterator() { impl = boost::make_shared<detail::DirectoryIteratorImpl>(); }; }; It seems like DirectoryIterator should be able to call boost::make_shared<DirectoryIteratorImpl>, because it is a friend of DirectoryIteratorImpl. However, this code fails to compile because the constructor for DirectoryIteratorImpl is private. Since this class is an internal implementation detail that clients of DirectoryIterator should never touch, it would be nice if I could keep the constructor private. Is this my fundamental misunderstanding around make_shared or do I need to mark some sort of boost piece as friend in order for the call to compile?

    Read the article

  • How to link to dynamic boost libs?

    - by testingmysql
    I compiled boost lib and got these. //Shared/dynamic link libraries 24/03/2010 11:25 PM 53,248 boost_thread-vc80-mt-1_42.dll 24/03/2010 11:25 PM 17,054 boost_thread-vc80-mt-1_42.lib 24/03/2010 11:25 PM 17,054 boost_thread-vc80-mt.lib 24/03/2010 11:25 PM 73,728 boost_thread-vc80-mt-gd-1_42.dll 24/03/2010 11:25 PM 17,214 boost_thread-vc80-mt-gd-1_42.lib 24/03/2010 11:25 PM 17,214 boost_thread-vc80-mt-gd.lib // Static libs... does not need any dlls 24/03/2010 11:25 PM 381,716 libboost_thread-vc80-mt-1_42.lib 24/03/2010 11:25 PM 381,716 libboost_thread-vc80-mt.lib 24/03/2010 11:25 PM 999,552 libboost_thread-vc80-mt-gd-1_42.lib 24/03/2010 11:25 PM 999,552 libboost_thread-vc80-mt-gd.lib 24/03/2010 11:25 PM 421,050 libboost_thread-vc80-mt-s-1_42.lib 24/03/2010 11:25 PM 421,050 libboost_thread-vc80-mt-s.lib 24/03/2010 11:25 PM 1,015,688 libboost_thread-vc80-mt-sgd-1_42.lib 24/03/2010 11:25 PM 1,015,688 libboost_thread-vc80-mt-sgd.lib In Visual Studio, I have written a test app using the boost thread library. Based on code generation settings it asks for these four libs only (like multithreading debug, multithreading, multithreading debug dll, and multithreading dll) 24/03/2010 11:25 PM 381,716 libboost_thread-vc80-mt-1_42.lib 24/03/2010 11:25 PM 381,716 libboost_thread-vc80-mt.lib 24/03/2010 11:25 PM 999,552 libboost_thread-vc80-mt-gd-1_42.lib 24/03/2010 11:25 PM 999,552 libboost_thread-vc80-mt-gd.lib 24/03/2010 11:25 PM 421,050 libboost_thread-vc80-mt-s-1_42.lib 24/03/2010 11:25 PM 421,050 libboost_thread-vc80-mt-s.lib 24/03/2010 11:25 PM 1,015,688 libboost_thread-vc80-mt-sgd-1_42.lib 24/03/2010 11:25 PM 1,015,688 libboost_thread-vc80-mt-sgd.lib Now my question is how can I link my app to the other 2 libs so that it uses the dlls? 24/03/2010 11:25 PM 53,248 boost_thread-vc80-mt-1_42.dll 24/03/2010 11:25 PM 17,054 boost_thread-vc80-mt-1_42.lib 24/03/2010 11:25 PM 17,054 boost_thread-vc80-mt.lib 24/03/2010 11:25 PM 73,728 boost_thread-vc80-mt-gd-1_42.dll 24/03/2010 11:25 PM 17,214 boost_thread-vc80-mt-gd-1_42.lib 24/03/2010 11:25 PM 17,214 boost_thread-vc80-mt-gd.lib Question 2. What does the g, s stands for?

    Read the article

  • Composite key syntax in Boost MultiIndex

    - by Sarah
    Even after studying the examples, I'm having trouble figuring out how to extract ranges using a composite key on a MultiIndex container. typedef multi_index_container< boost::shared_ptr< Host >, indexed_by< hashed_unique< const_mem_fun<Host,int,&Host::getID> >, // ID index ordered_non_unique< const_mem_fun<Host,int,&Host::getAgeInY> >, // Age index ordered_non_unique< const_mem_fun<Host,int,&Host::getHousehold> >, // Household index ordered_non_unique< // Age & eligibility status index composite_key< Host, const_mem_fun<Host,int,&Host::getAgeInY>, const_mem_fun<Host,bool,&Host::isPaired> > > > // end indexed_by > HostContainer; My goal is to get an iterator pointing to the first of the subset of elements in HostContainer hmap that has age partnerAge and returns false to Host::isPaired(): std::pair< hmap::iterator,hmap::iterator > pit = hmap.equal_range(boost::make_tuple( partnerAge, false ) ); I think this is very wrong. How/Where do I specify the iterator index (which should be 3 for age & eligibility)? I will include other composite keys in the future. What exactly are the two iterators in std::pair? (I'm copying syntax from an example that I don't understand.) I would ideally use std::count to calculate the number of elements of age partnerAge that are eligible (return false to Host::isPaired()). What is the syntax for extracting the sorted index that meets these requirements? I'm obviously still learning C++ syntax. Thanks in advance for any help.

    Read the article

  • Mocking with Boost::Test

    - by Billy ONeal
    Hello everyone :) I'm using the Boost::Test library for unit testing, and I've in general been hacking up my own mocking solutions that look something like this: //In header for clients struct RealFindFirstFile { static HANDLE FindFirst(LPCWSTR lpFileName, LPWIN32_FIND_DATAW lpFindFileData) { return FindFirstFile(lpFileName, lpFindFileData); }; }; template <typename FirstFile_T = RealFindFirstFile> class DirectoryIterator { //.. Implementation } //In unit tests (cpp) #define THE_ANSWER_TO_LIFE_THE_UNIVERSE_AND_EVERYTHING 42 struct FakeFindFirstFile { static HANDLE FindFirst(LPCWSTR lpFileName, LPWIN32_FIND_DATAW lpFindFileData) { return THE_ANSWER_TO_LIFE_THE_UNIVERSE_AND_EVERYTHING; }; }; BOOST_AUTO_TEST_CASE( MyTest ) { DirectoryIterator<FakeFindFirstFile> LookMaImMocked; //Test } I've grown frustrated with this because it requires that I implement almost everything as a template, and it is a lot of boilerplate code to achieve what I'm looking for. Is there a good method of mocking up code using Boost::Test over my Ad-hoc method? I've seen several people recommend Google Mock, but it requires a lot of ugly hacks if your functions are not virtual, which I would like to avoid. Oh: One last thing. I don't need assertions that a particular piece of code was called. I simply need to be able to inject data that would normally be returned by Windows API functions.

    Read the article

  • Boost class/struct serialization to byte array

    - by Dave18
    does boost library provide functions to pack the class/struct data into a byte array to shorten the length of serialized data? Currently i'm using stringstream to get the serialized data, for example - struct data { std::string s1; std::string s2; int i; }; template <typename Archive> void serialize(Archive &ar, data &d, const unsigned int version) { ar & d.s1; ar & d.s2; ar & d.i; } int main() { data d; d.s1 = "This is my first string"; d.s2 = "This is my second string"; d.i = 10000; std::stringstream archive_stream; boost::archive::text_oarchive archive(archive_stream); archive.operator <<(d); } How would i use a byte array instead of stringstream for data?

    Read the article

  • How to pass parameters to manage_shared_memory.construct() in Boost.Interprocess

    - by recipriversexclusion
    I've stared at the Boost.Interprocess documentation for hours but still haven't been able to figure this out. In the doc, they have an example of creating a vector in shared memory like so: //Define an STL compatible allocator of ints that allocates from the managed_shared_memory. //This allocator will allow placing containers in the segment typedef allocator<int, managed_shared_memory::segment_manager> ShmemAllocator; //Alias a vector that uses the previous STL-like allocator so that allocates //its values from the segment typedef vector<int, ShmemAllocator> MyVector; int main(int argc, char *argv[]) { //Create a new segment with given name and size managed_shared_memory segment(create_only, "MySharedMemory", 65536); //Initialize shared memory STL-compatible allocator const ShmemAllocator alloc_inst (segment.get_segment_manager()); //Construct a vector named "MyVector" in shared memory with argument alloc_inst MyVector *myvector = segment.construct<MyVector>("MyVector")(alloc_inst); Now, I understand this. What I'm stuck is how to pass a second parameter to segment.construct() to specify the number of elements. The interprocess document gives the prototype for construct() as MyType *ptr = managed_memory_segment.construct<MyType>("Name") (par1, par2...); but when I try MyVector *myvector = segment.construct<MyVector>("MyVector")(100, alloc_inst); I get compilation errors. My questions are: Who actually gets passed the parameters par1, par2 from segment.construct, the constructor of the object, e.g. vector? My understanding is that the template allocator parameter is being passed. Is that correct? How can I add another parameter, in addition to alloc_inst that is required by the constructor of the object being created in shared memory? There's very little information other than the terse Boost docs on this.

    Read the article

  • boost::shared_ptr in Objective-C++

    - by John Smith
    This is a better understanding of a question I had earlier. I have the following Objective-C++ object @interface OCPP { MyCppobj * cppobj; } @end @implementation OCPP -(OCPP *) init { cppobj = new MyCppobj; } @end Then I create a completely differently obj which needs to use cppobj in a boost::shared_ptr (I have no choice in this matter, it's part of a huge library which I cannot change) @interface NOBJ -(void) use_cppobj_as_shared_ptr { //get an OCPP obj called occ from somewhere.. //troubling line here } @end I have tried the following and that failed: I tried synthesising cppobj. Then I created a shared_ptr in "troubling line" in the following way: MyCppobj * cpp = [occ cppobj]; bsp = boost::shared_ptr<MyCppobj>(cpp); It works fine first time around. Then I destroy the NOBJ and recreate it. When I for cppobj it's gone. Presumably shared_ptr decided it's no longer needed and did away with it. So I need help. How can I keep cppobj alive? Is there a way to destroy bsp (or it's reference to cppobj) without destroying cppobj?

    Read the article

  • Return Double from Boost thread

    - by Benedikt Wutzi
    Hi I have an Boost thread which should return a double. The function looks like this: void analyser::findup(const double startwl, const double max, double &myret){ this->data.begin(); for(int i = (int)data.size() ; i >= 0;i--){ if(this->data[i].lambda > startwl){ if(this->data[i].db >= (max-30)) { myret = this->data[i+1].lambda; std::cout <<"in thread " << myret << std::endl; return; } } } } this function is called by another function: void analyser::start_find_up(const double startwl, const double max){ double tmp = -42.0; boost::thread up(&analyser::findup,*this, startwl,max,tmp); std::cout << "before join " << tmp << std::endl; up.join(); std::cout << "after join " << tmp << std::endl; } Anyway I've tried and googled almost anything but i can't get it to return a value. The output looks like this right now. before join -42 in thread 843.487 after join -42 Thanks for any help.

    Read the article

  • Noob boost::bind member function callback question

    - by shaz
    #include <boost/bind.hpp> #include <iostream> using namespace std; using boost::bind; class A { public: void print(string &s) { cout << s.c_str() << endl; } }; typedef void (*callback)(); class B { public: void set_callback(callback cb) { m_cb = cb; } void do_callback() { m_cb(); } private: callback m_cb; }; void main() { A a; B b; string s("message"); b.set_callback(bind(A::print, &a, s)); b.do_callback(); } So what I'm trying to do is to have the print method of A stream "message" to cout when b's callback is activated. I'm getting an unexpected number of arguments error from msvc10. I'm sure this is super noob basic and I'm sorry in advance.

    Read the article

  • Using boost asio for pub/sub style tcp in a game loop

    - by unohoo
    I have been reading through the boost asio documentation for a couple of hours now, and while I think the documentation is really great, I am still left a bit confused on how to implement the system that I need. I have to stream info, from a game engine, to a list of computers over tcp. One snag is that, unlike traditional pub/sub, the computer that does the distribution of info is actually the computer that has to connect to the subscribers as well (instead of the subscribers registering with the publisher). This is done via a config file - a list of ip's/ports along with the data that they each require. The subscribers listen, but do not know the ip of the publisher. (As a side note, I'm quite new to network programming, so maybe I'm missing something .. but it's strange that I do not find much information regarding this style of "inverted" client-server model..) I am looking for suggestions for the implementation of such a system using boost asio. Of course I have to integrate the networking into an already existing engine, so with regards to that: What would be a good way to handle messages being sent to multiple computers every frame? Use async_write, call io_service.run and then reset every frame? Would having io_service.run have its own thread be better? Or should I just use threads and use blocking writes?

    Read the article

  • C++ destructor problem with boost::scoped_ptr

    - by bb-generation
    I have a question about the following code: #include <iostream> #include <boost/scoped_ptr.hpp> class Interface { }; class A : public Interface { public: A() { std::cout << "A()" << std::endl; } virtual ~A() { std::cout << "~A()" << std::endl; } }; Interface* get_a() { A* a = new A; return a; } int main() { { std::cout << "1" << std::endl; boost::scoped_ptr<Interface> x(get_a()); std::cout << "2" << std::endl; } std::cout << "3" << std::endl; } It creates the following output: 1 A() 2 3 As you can see, it doesn't call the destructor of A. The only way I see to get the destructor of A being called, is to add a destructor for the Interface class like this: virtual ~Interface() { } But I really want to avoid any Implementation in my Interface class and virtual ~Interface() = 0; doesn't work (produces some linker errors complaining about a non existing implementation of ~Interface(). So my question is: What do I have to change in order to make the destructor being called, but (if possible) leave the Interface as an Interface (only abstract methods).

    Read the article

  • Tips for submitting a library to Boost?

    - by AraK
    Hi everyone, Summer is coming, and a group of friends and I are getting ready for it :) We decided to build a compile-time Arbitrary precision Unsigned Integers. We would like to provide a set of integers algorithms(functions) with the library. We have seen a number of requests for such a library(SoC2010, C++0x Standard Library wishlist). Also, a regular run-time bigint is requested usually with that, but we don't want to go into the hassle of memory management. The idea came to me from a library called TTMath, unfortunately this library works only on specific platforms because Assembly was used extensively in the library. We would like to write a standard library, depending on the C++ standard library and Boost. Also, we would like to use the available C++0x facilities in current compilers like user-defined literals and others. This would technically make the library non-standard for a while, but we think that it is a matter of time the new standards will be official. Your hints on the whole process including design, implementation, documentation, maintainable of the library are more than welcom. We are a group of students and fresh graduates who are looking for something interesting in the summer, but we see that Boost is full of gurus and we don't want to forget something too obvious. We are communicating on-line, so there is no shared white-boards :( Thanks,

    Read the article

  • Failed at linking C++ [undefined reference boost::filesystem3 ... ]

    - by Pphax
    i'm having some troubles compiling my work, i'm using ubuntu with g++! i get a lot of these messages: undefined reference to `boost::filesystem3::directory_entry::m_get_status(boost::system::error_code*) const' undefined reference to `boost::filesystem3::path::extension() const' undefined reference to `boost::filesystem3::path::filename() const' undefined reference to `boost::filesystem3::path::filename() const' (etc...) I've searched and found maaany answers but none of those work for me. [...] -lboost_system (/usr/lib/gcc/i686-linux-gnu/4.4.5/../../../../lib/libboost_system.so) -lboost_filesystem (/usr/lib/gcc/i686-linux-gnu/4.4.5/../../../../lib/libboost_filesystem.so) [...] (when linking it shows those two libraries, i'm guessing the error is related to the second one. hax@lap:~$ locate libboost_filesystem.so /home/hax/boost_1_47_0/bin.v2/libs/filesystem/build/gcc-4.4.5/release/threading-multi/libboost_filesystem.so.1.47.0 /home/hax/boost_1_47_0/stage/lib/libboost_filesystem.so /home/hax/boost_1_47_0/stage/lib/libboost_filesystem.so.1.47.0 /usr/lib/libboost_filesystem.so /usr/lib/libboost_filesystem.so.1.42.0 /usr/local/lib/libboost_filesystem.so /usr/local/lib/libboost_filesystem.so.1.47.0 this is the related line on my makefile: -L. -L../bncsutil/src/bncsutil/ -L../StormLib/stormlib/ -L../boost/lib/ -lbncsutil -lpthread -ldl -lz -lStorm -lmysqlclient_r -lboost_date_time -lboost_thread -lboost_system -lboost_filesystem -Wl -t I tried pointing with -L several different places where i saw filesystem.so was located but it didn't work! Can anyone see the problem in those lines? if you need me to put some extra data i'll do it, i'm not seeing the problem :( Thanks :)

    Read the article

  • remove versioning on boost xml serialization

    - by cppanda
    hi, i just can't find a way to remove the version tracking from the boost xmlarchives. example <Settings class_id="0" tracking_level="0" version="1"> <px class_id="1" tracking_level="1" version="0" object_id="_0"> <TestInt>3</TestInt> <Resolution class_id="2" tracking_level="0" version="0"> <x>800</x> <y>600</y> </Resolution> <SomeStuff>0</SomeStuff> </px> </Settings> I want to get ride of the class_id="0" tracking_level="0" version="1" stuff, because for in this case i just don't need it and want a simple clean config like file code void serialize(Archive & ar, const unsigned int version) { ar & make_nvp("TestInt", TestInt); ar & make_nvp("Resolution", resolution); ar & make_nvp("SomeStuff", SomeStuff); } i found boost::serialization::track_never, but nowhere to use it

    Read the article

  • How to setup directories in Visual Studio when using boost?

    - by Rich
    Hi, I have introduced boost to our code base, on my machine I created a boost directory called Thirdparty.Boost and added that as an additional include directory in my Visual Studio setting, all is fine. However I now want to check in my changes, so the rest of the team can get them. Inorder to build the code they would need to setup boost as I have (problem number 1). In addition we have a build server, which will need changing (problem 2). I have a way of distributing boost to everyone including the build server, so that's not a problem I need a way of referring to the boost directory without changing the default settings in Visual Studio. Why don't you change it on a project level I hear you cry? The solution has over 200 projects, which would require a lot of changes. I just wondered if there was another way? Cheers Rich

    Read the article

  • Is Boost.Tuple compatible with C++0x variadic templates ?

    - by Thomas Petit
    Hi, I was playing around with variadic templates (gcc 4.5) and hit this problem : template <typename... Args> boost::tuple<Args...> my_make_tuple(Args... args) { return boost::tuple<Args...>(args...); } int main (void) { boost::tuple<int, char> t = my_make_tuple(8, 'c'); } GCC error message : sorry, unimplemented: cannot expand 'Arg ...' into a fixed-length argument list In function 'int my_make_tuple(Arg ...)' If I replace every occurrence of boost::tuple by std::tuple, it compiles fine. Is there a problem in boost tuple implementation ? Or is this a gcc bug ? I must stick with Boost.Tuple for now. Do you know any workaround ? Thanks.

    Read the article

< Previous Page | 10 11 12 13 14 15 16 17 18 19 20 21  | Next Page >