- Boost-users - lists.boost.org

Documentation Suggestion.
by Minkoo Seo 30 Aug '06

30 Aug '06

Hi. I enjoy boost library these days, and really appreciate it. However, sometimes, I found it very hard to even compile a program that uses boost libraries. As an example, serialization library needs libboost_serialization.so. Hence, I have to compile a program like the following: g++ -lboost_serializatino foo.cpp Nevertheless, AFAIK, nowhere in the document, how to compile a program is not presented. Some of the docs even mention the header file. I don't know why this is so common in the documentation of boost, and I think mentioning how to compile will be very useful for users. So, please, please put some 'how to compile' section especially if the library needs -lxxx_library flag. I really thank boost developers' time and efforts. Sincerely, Minkoo Seo -- View this message in context: http://www.nabble.com/Documentation-Suggestion.-tf2189683.html#a6058183 Sent from the Boost - Users forum at Nabble.com.

3 4

[iostreams] Auto flush required
by Roman Perepelitsa 30 Aug '06

30 Aug '06

Hi. I've implemented output stream using boost.iostreams library. It requires call to flush in order to actually write data to sink. But I want to flush data to sink automatically after I write any object with operator<<. struct stream_sink : public boost::iostreams::sink { stream_sink(std::ostream & strm) : strm_(strm) {} std::streamsize write(const char * s, std::streamsize n) { if (strm_.write(s, n)) return n; else throw std::runtime_error("error"); } private: std::ostream & strm_; }; std::stringstream s; boost::iostreams::stream<stream_sink> strm(s); strm << "abc"; strm.flush(); std::cout << "String: '" << s.str() << "'" << std::endl; This test outputs "String: 'abc'" but if you comment strm.flush() then it outputs "String: ''". I want my stream to work like the stringstream does (it does not require call to flush). std::stringstream strm; strm << "abc"; std::string s; strm >> s; std::cout << "String: '" << s << "'" << std::endl; It outputs "String: 'abc'". I need it to create stream which can be used to convert wide char data (wchar_t *) to multibyte (char *). It can be used this way: // wcstombsstream is wostream and istream // "utf-8" is multibyte encoding wcstombsstream strm("utf-8"); strm << L"abc"; std::string s; strm >> s; This code does not work at the moment because it requires to call strm.flush() before reading from stream. Thus I need some auto flush facility :-) Thank you for any help, Roman Perepelitsa.

1 1

Boost Serialization lib:Problem in object tracking and Exporting classes
by mahesh prabhu 30 Aug '06

30 Aug '06

Hi, I'm new to Boost Serialization library . I wanted to use the library for some of my full fledge so I thought I might as well test out the features that the library documentation claimed it had . Many of the basic features did work but some of the advanced features that I wanted most like object tracking and serializing derived objects through a pointer to the base class gave me some problems . I faced the following two problems ( in order of decreasing priority ): (1) Object Serialization : I have 4 pointers to 3 objects of the same class , On deserializing these the pointers are mixed up . If while writing out pointers 1 & 2 pointed to objects 1 and 2 , and pointers 3 and 4 pointed to objects 1 and 3 then on restoration pointers 1 & 2 pointed to objects 1 & 2 but pointers 3 & 4 pointed to objects 2 and 3 instead of 1 and 3 . (2) Writing derived objects through pointer to base class : This feature worked if I did a registration through ar.register_type(static_cast<gps_moreAcc1 *>(NULL)) but did not work if I did BOOST_CLASS_EXPORT( gps_moreAcc1) . I get a "DoneAbort" message i.e a core dump . I don't know too much about templates , so I was not able to debug the problem . I have attached a program which exhibit both problems , but the code that causes the second problem is commented out . This can be reproduced by just commenting out ar.register_type calls and uncommenting BOOST_CLASS_EXPORT macros . I have added more comments in the source file . Please let me know if I am doing something wrong or if these are know problems with the serialization library or if the serialization library cannot be handled by serialization . +Mahesh

1 0

Re: [Boost-users] timed_mutex::scoped_timed_lock
by Andrew Wingorodov 30 Aug '06

30 Aug '06

10x, sorry for my english > I am not sure what this delay function does. this function from ${BOOSTDIR}/libs/thread/test/util.inl > scoped_time_lock takes boost::xtime or bool. 4 here is interpreted as > "true", not as time. oops. okey, the new version of my cool program (how i can see event of timeout?) are ///------- COOLPRG timed_mutex mut; #define TIMEOUT 1 #define AUTHTIME 4 void timeout () { timed_mutex::scoped_timed_lock lk (mut, delay(TIMEOUT)); if ( lk.locked ()) cout << "how i can see event of timeout?" << endl; else cout << "okey!" << endl; } std::auto_ptr<timed_mutex::scoped_timed_lock> lock; void auth () { cout << "auth" << endl; boost::thread::sleep ( delay(AUTHTIME)); cout << "auth end" << endl; lock->unlock(); } int main () { lock = (std::auto_ptr<timed_mutex::scoped_timed_lock>) new timed_mutex::scoped_timed_lock (mut,10); thread_group run; run.create_thread ( &auth); run.create_thread ( &timeout); run.create_thread ( &timeout); run.create_thread ( &timeout); run.join_all(); if ( !lock->locked ()) cout << "unlock!" << endl; } ///EOF And result (how i can see event of timeout?): andr@andr> ./coolprg auth okey! okey! okey! auth end unlock! andr@andr> -- andr.ru

1 0

[BGL] access order in outEdgeList containers ?
by Manuel Yguel 30 Aug '06

30 Aug '06

Hello, Does the order of the add_edge() calls for an adjacency_list graph with vecS or listS OutEdgeList containers defines the access order through out_edge_iterators ? If true, how to define some add_edge_after() method with the same semantic than the insert_after method defined for slist in http://www.sgi.com/tech/stl/Slist.html Thanks a lot. P.S. thanks Doug, for your answer concerning graph properties.

1 0

[regex] - Don't give wrong documentation in boost site itself.
by kiran 30 Aug '06

30 Aug '06

Hi This mail is to say that wrong documentation in a site like boost in no worser than no documentation at all. http://www.boost.org/libs/regex/doc/syntax_perl.html In this the second line saying that "Perl regular expressions are the default behavior in Boost.Regex" and then when using boost::regex i found that "dot" in perl and boost::regex have different meaning by default. I am not speaking aganist the guy who made the documentation. He did a great job by creating the doc which help newbies like me to great extent. The aim of this mail is to get the doc corrected. Could anyone tell what are the differences between perl regex and boost::regex by default . Thanks Kiran.

1 0

[regex] - Doubts of a regex beginner on boost::regex getting aborted.
by kiran 30 Aug '06

30 Aug '06

Hi, I have 2 doubts 1) Why is regex expr("Resurfacing.*?Home", boost::regex::icase) getting aborted. ? 2) How to make boost regex behave like perl regex ? I mean "dot" in boost::regex is taking new lines as well. How to prevent this ? Is "boost::regex::perl" flag the answer to the question ? Thanks in advance Kiran.

2 1

Mismatch and regex newbie problem still problem
by david v 30 Aug '06

30 Aug '06

So to sum-up. If the regex i'm looking for is "testing" and the string to search the regex for is "tastung" (obviously this is a short example but i'm dealing with more complex regular expressions. how can i get the number of mismatches. Basically the output of the program would tell me: >2 mismatches found in string "tastung" at position 2 (a) and 5(u). I have tried the iterator above but does not seem to do what i want??? std::string::const_iterator start = in.begin(); std::string::const_iterator end = in.end(); boost::sregex_token_iterator i(start, end, *regex, -1), j; while(i != j) { cout << *i << endl; }

2 2

timed_mutex::scoped_timed_lock
by Andrew Wingorodov 30 Aug '06

30 Aug '06

Why result of my program such? auth() sleeps one second, and I have established four in timeout. I thought, that will be "okey". why timeout? # ./a auth auth end timeout! timeout! timeout! # ///----------------------------- static timed_mutex mut; void timeout () { timed_mutex::scoped_timed_lock lk (mut,4); if ( lk.locked ()) cout << "timeout!" << endl; else cout << "okey!" << endl; } timed_mutex::scoped_timed_lock *lock; void auth () { cout << "auth" << endl; boost::thread::sleep ( delay(1)); cout << "auth end" << endl; delete lock; } int main () { lock = new timed_mutex::scoped_timed_lock (mut,10); thread_group run; run.create_thread ( &auth); run.create_thread ( &timeout); run.create_thread ( &timeout); run.create_thread ( &timeout); run.join_all(); } -- andr.ru

3 2

reference_wrapper question
by Ovanes Markarian 29 Aug '06

29 Aug '06

Hello, I have a small problem. Some of my data holders have dereference operator which returns refernces to contained types. The problem I face here is that I have some noncopyable types and the dereference operator fails to compile. I thought using a reference_wrapper class could do the trick, but then I can not call functions on the contained type without explict casting, regardless operator T&() const defined. here is a small test app to simulate the behavior: #include <boost/ref.hpp> #include <boost/array.hpp> #include <boost/utility.hpp> using namespace std; template<class T> struct test { test() : ptr(new T()) {} ~test() { delete ptr; } boost::reference_wrapper<T> operator*() { return boost::ref(*ptr); } T* ptr; }; template<class T, int N> class array : public boost::array<T,N> , private boost::noncopyable {}; int main(int argc, char* argv[]) { test< array<int,10> > t; int size = (static_cast<array<int,10>&>(*t)).size(); } Is there any way to call the last line like: int size = (*t).size(); operator ->() would work just fine, but I need to support both. Thanks for help, Ovanes Markarian

1 0