- Boost-users - lists.boost.org

[1.35] configuring for 'complete'
by Christoph Duelli 04 Apr '08

04 Apr '08

bjam has the option --build-type=complete Is it somehow possible (other than manually editing the Makefile) to specify this option when calling configure? (I'm asking because I'd like to script the build of Boost; and so far it only builds in release mode. I'd like to avoid to edit Makefile on any machine I need to build Boost on.) And: when I am using the (non-)multi-threaded boost libs, do I need to #define a variable so that the header-only libs are thread-safe too? Thank you, best regards Christoph

3 4

[GIL] Default constructing pixels??
by John Femiani 04 Apr '08

04 Apr '08

I am used to explicitly invoking the default constructor to initialize values to 'zero', but that does not seem to work with GIL. In the following code: template<class View> void do_something(View const& img){ typedef typename View::value_type pixel_type; pixel_type black = pixel_type(); //I expect zero, I guess this may not always be black... //..... } The value held in 'black' turns out to be 205 (I think it is acting like uninitialized stack data). Is this the intended behavior? I have a slew of code that used to work this way and somehow now it does not - did this behavior change recently? In the meantime I just made a struct named black with a typecast operator enabled for gil pixels. -- John

2 2

unordered_map memory footprint
by Robert Dailey 03 Apr '08

03 Apr '08

Hey guys, Hash tables are new to me since they have never been part of the C++ standard template library, and having started C++ as my first language I have never been exposed to them. Ever since unordered_map appeared in the Boost library, I've been trying to understand how they function in terms of functionality, memory, and performance. A major source of information has been from Wikipedia, specifically this page<http://en.wikipedia.org/wiki/Hash_table>. Below I have outlined a few questions concerning boost::unordered_map: 1. From what I've read, the performance of a hash table is O(1), however it mentions that depending on specific situations it may be O(n). Is this true with unordered map? If so, what conditions would force it to maintain O(n) complexity? 2. What is the memory footprint of an unordered_map? From my research, hash tables seem to be a simple wrapper over an array in that the size of the "internal array" in the hash table will contain as many elements as the largest hash value generated by one of the key elements. So if I put two elements in my hash table and the generated keys are 200 and then 1,000,000, then the hash table would be 1,000,000 elements in size (which is at least a megabyte of memory in the most optimistic scenario). This logic is rather ridiculous and impractical, so I am pretty sure I'm misunderstanding how the memory mechanics work under the hood for an unordered_map. Those are the biggest questions I have now and if I think of more I'll most certainly be providing follow-up posts. I thank everyone in advance for reading and helping.

4 6

[iostreams] newline_filter problem
by Soren Soe 03 Apr '08

03 Apr '08

Hi, It appears boost::iostreams::newline_filter(posix) converts all CRs into LFs rather than just converting the sequence CRLF. In other words: "\rtext\r\r\n" is converted to "\ntext\n\n" when I would have expected "\rtext\r\n". Indeed the latter is what dos2unix does on my linux system. Is this a bug in the newline_filter? Thanks, Soren Soe

2 1

[lexical_cast] for user type extensions.
by Balasubramanyam, Shivakumar 03 Apr '08

03 Apr '08

Hi, I need conversion from strings to integers. But the strings could also be hex numbers. So example is, typedef unsigned long hexType; hexType value; lexical_cast<HexType>("0xAABBCCDD"); How do we extend the boost::lexical_cast to support casting user defined types? I can perform this by specialization but those would need to be in boost namespace which does not seem to be a recommended way to add customizations. Thanks, Shiva

3 3

Datetime: time_of_day and time_range concept
by Jean-Pierre Bergamin 03 Apr '08

03 Apr '08

Hello everyone While implementing a job scheduler that has to start jobs at specific times (i.e. daily at 08:15) and during specific time ranges (every Monday between 16:20 and 18:00), the need for two new concepts that represents a time (detached from a date) and a time range between two time_of_day came up. While time_duration almost fits these needs, there are some things that we had to take into account: - A time_of_day must be normalized to only allow hours >= 0 && hours < 24 - Only the substract operator makes sense to calculate the duration between to times (15:30 - 12:00 = 3 hours and 30 minutes). 12:20 + 15:32 does not make sense. The implicit conversion to time_duration also allows the usage of time_of_day to create ptime instances (which is IMHO more "understandable" then using time_duration): ptime some_time(date(2008, 4, 1), time_of_day(14, 20, 0)); The implemented classes time_of_day (derived from time_duration) and time_range are very lightweight and allow the user to do things like: BOOST_AUTO_TEST_CASE( time_of_day_test2 ) { time_of_day t1(15,0,0); time_of_day t2(40,0,0); // normalized to 16:00:00 BOOST_MESSAGE("t1 " << t1); // t1 15:00:00 BOOST_MESSAGE("t2 " << t2); // t2 16:00:00 t2 += hours(3) + minutes(30); BOOST_MESSAGE("t1 " << t1); // t1 15:00:00 BOOST_MESSAGE("t2 " << t2); // t2 19:30:00 time_duration d1(t2 - t1); // Duration from 15:00 to 19:30 = 4:30 hour time_duration d2(t1 - t2); // Duration from 19:30 to 15:00 = 19:30 hours BOOST_MESSAGE("d1 " << d1); // d1 04:30:00 BOOST_MESSAGE("d2 " << d2); // d2 19:30:00 time_range afternoon(t1, t2); // From 15:00 to 19:30 time_range over_night(t2, t1); // From 19:30 to 15:00 (over midnight) ptime p1(date(2008, 4, 1), time_of_day(14, 20, 0)); ptime p2(date(2008, 4, 1), time_of_day(16, 40, 0)); BOOST_CHECK(afternoon.contains(p1) == false); BOOST_CHECK(afternoon.contains(p2) == true); BOOST_CHECK(over_night.contains(p1) == true); BOOST_CHECK(over_night.contains(p2) == false); time_range tr1(time_of_day(15, 0, 0), time_of_day(18, 0, 0)); // 15:00 to 18:00 time_range tr2(time_of_day(16, 0, 0), time_of_day(17, 0, 0)); // 16:00 to 17:00 time_range tr3(time_of_day(17, 0, 0), time_of_day(19, 0, 0)); // 17:00 to 19:00 BOOST_CHECK(tr1.contains(tr2) == true); BOOST_CHECK(tr2.contains(tr1) == false); BOOST_CHECK(tr2.contains(tr3) == false); } I think these concepts would fit well into the date time lib. What do you think? Regards James

2 1

Re: [Boost-users] serialization, 1.34/1.35 difference: serializing a vector
by Filip Peters 03 Apr '08

03 Apr '08

I looked a bit deeper into the code and found this: in 1.34.1, a vector would use the code from collections_save_imp.hpp, which contains the following: unsigned int count = s.size(); ar << BOOST_SERIALIZATION_NVP(count); if(3 < ar.get_library_version()){ const unsigned int item_version = version< BOOST_DEDUCED_TYPENAME Container::value_type >::value; ar << BOOST_SERIALIZATION_NVP(item_version); } in 1.35.0, there is an "optimized" version in vector.hpp, const collection_size_type count(t.size()); ar << BOOST_SERIALIZATION_NVP(count); if (!t.empty()) ar << make_array(detail::get_data(t),t.size()); without the item_version. Loading a vector of a type like int or double, created in 1.34 won't work in 1.35. Filip Peters. Haven't used boost's trac before, but I'll enter it tomorrow. In the meantime, adding item_version into the optimized versions seems to solve my problem, except I have to make sure it doesn't break in the next version. ----- Original Message ---- From: Robert Ramey <ramey(a)rrsd.com> To: boost-users(a)lists.boost.org Sent: Wednesday, April 2, 2008 6:30:01 PM Subject: Re: [Boost-users] serialization, 1.34/1.35 difference: serializing a vector Hmm I don't recal where that might have come from. Open a TRAK item. Robert Ramey "Filip Peters" <filippeters(a)yahoo.com> wrote in message news:859190.51711.qm@web30705.mail.mud.yahoo.com... I just tried serializing a simple vector, std::vector<unsigned int> vectortest; If I serialize out in 1.34, I can't read it in in 1.35. xml result from 1.35: <vectortest> <count>3</count> <item>1</item> <item>2</item> <item>3</item> </vectortest> from 1.34: <vectortest> <count>3</count> <item_version>0</item_version> <item>1</item> &n bsp; <item>2</item> <item>3</item> </vectortest> Notice the item_version, this causes an exception when reading it in with 1.35.... I need a way to get the behavior from 1.34 back.... Otherwise my files aren't readable anymore. small testprogram: #include <boost/config.hpp> #include <boost/archive/xml_oarchive.hpp> #include <boost/archive/xml_iarchive.hpp> //#include <boost/archive/binary_iarchive.hpp> //#include <boost/archive/binary_oarchive.hpp> #include <boost/serialization/vector.hpp> #include <boost/serialization/export.hpp> #include <fstream> #include <iomanip> int _tmain(int argc, _TCHAR* argv[]) { std::vector<unsigned int> testvector; testvector.push_back(1.); testvector.push_back (2.); testvector.push_back(3.); //this in 1.34 std::ofstream ofs( "testing2.txt", std::ios::binary ) ; boost::archive::xml_oarchive oa(ofs); oa << boost::serialization::make_nvp("vectortest", testvector); ofs.close(); //do this in 1.35 try{ std::ifstream ofs( "testing2.txt", std::ios::binary ) ; boost::archive::xml_iarchive ia(ofs); ia >> boost::serialization::make_nvp("vectortest", testvector2); ofs.close(); } catch(boost::archi ve::archive_exception testexception){ //reading a file from 1.34, will end up here std::string error = testexception.what(); } return 0; } You rock. That's why Blockbuster's offering you one month of Blockbuster Total Access, No Cost. _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users ____________________________________________________________________________________ You rock. That's why Blockbuster's offering you one month of Blockbuster Total Access, No Cost. http://tc.deals.yahoo.com/tc/blockbuster/text5.com

2 1

boost::iequals bug in 1.35?
by L Pocaille 03 Apr '08

03 Apr '08

3 5

asio::async_read and buffer handling
by Axel 03 Apr '08

03 Apr '08

Hello I m a new Boost user, (and a little less but still new C++ developper) and I d like to find some advices about how to manage network communications. I m writing a daemon and I use boost::asio::async_read_until() to fill a boost::asio::streambuf buffer. Here's the corresponding handler : void Session::handleRecv(const boost::system::error_code& error, size_t size) { if(error.value() == 0 && size > 0) { size -= separator.size(); istream command_stream(&buffer_); char * command_chr = new char[size]; command_stream.read(command_chr, size); string command = command_chr; cout << "COMMAND " << command << endl; buffer_.consume(separator.size()); //doSomethingWithTheCommand(command); this->start(); } } I have some questions regarding this code. If I don't consume the full buffer_ , as soon as the this->start() is called, the handlerRecv() is called again. Is this a normal behaviour ? Is there a better way to extract the command from the buffer_ variable ? Finally how could I use an iterator of buffer_ without using an external class (like it's done in http://asio.sourceforge.net/boost_asio_1_0_0/libs/asio/doc/html/boost_asio/… , example I didn't fully understood) Thanks in advance.

1 0

[Boost.Spirit] how to extend a grammar
by dariomt＠gmail.com 03 Apr '08

03 Apr '08

Lets say I have a simple calculator grammar like the one defined in trunk\libs\spirit\example\fundamental\calc_plain.cpp Now I would like to extend that grammar e.g. add support for sqrt() One way would be modifying the existing grammar class (patch provided) But what I would like is to create a new class that uses or inherits from the existing grammar class to provide the extended behavior. That is, I would need both the calculator and the extended_calculator classes available, and without code duplication. ¿Is this possible?

1 0