September 2009 - Boost-users

[boost][shared_ptr] Trying to make a thread safe Copy-on-write class
by Haledor wow 28 Sep '09

28 Sep '09

Hi, Im trying to make a thread safe Copy-on-write class, but i am unsure how. I want a template class that stores a std::set<T> and i want to be able to add/remove elements while using iterators at the same time without conflicts. I would like to avoid making copies of the entire set each time i want to iterate over the elements. I have tried to use shared_ptr.unique() to decide when to create a new copy of the set but i have been unsuccessful since shared_ptr.use_count() stays at the value 1. I am doing something wrong. I have tried to use something like this as my add function: mutable boost::mutex _mutex; typedef std::set<T> Set; typedef boost::shared_ptr<Set> SetPtr; void add(T element) { boost::unique_lock<boost::mutex> lock(_mutex); if (!(_mySet.unique() || _mySet == 0)) { Set* tmp = _mySet.get(); _mySet = SetPtr(new Set(*tmp)); } _mySet->insert(element); } My thought was that each time an iterator accesses the set the count would increase. Could i make it so (should i)? Any other ideas about a thread safe implementation of a set of elements? I hope someone can help me. Thanks.

1 0

[Asio] problem with deadline_timer
by moffet 28 Sep '09

28 Sep '09

This is a small program using deadline_timer with expiration time 1 millisecond. http://snipt.org/nKm When I run it over and over it will evantually stuck lie this: http://img39.imageshack.us/img39/1292/handh.jpg and I don't know why. The io_service.run() is started in another thread. The main thread is waiting for pressing 'any key', the I call post() method to cancel() the timer. But sometimes as you can see on the screenshot, cancel() is invoked, but the timer continues to run. It seems that run() method doesn't invoke the timer handler with boost::asio::error::operation_aborted error. I've noticed this effect will be encouraged by high CPU usage. Vista 64bit, VC++ 2008 Express, Boost 1.40.0 Why this is happening?

2 1

Asio uses select instead of kqueue by default on FreeBSD?
by Lych 28 Sep '09

28 Sep '09

Hi, It seems so in v1.39 and 1.40, without the macro BOOST_ASIO_DISABLE_KQUEUE defined. I've found it's due to a line in the file [boost/trunk/boost/asio/detail/kqueue_reactor_fwd.hpp]: ------------------------------------------------------------- #if defined(__MACH__) && defined(__APPLE__) // Define this to indicate that epoll is supported on the target platform. #define BOOST_ASIO_HAS_KQUEUE 1 ------------------------------------------------------------- That only checks whether it is a MacOS system to enable kqueue, but doesn't care any of FreeBSD. One said it's because some issues are not yet fixed on use of kqueue on FreeBSD, but this seems to exist for quite long. Is there any more information? Thanks

1 0

[fusion] problem with make_map
by Jean-Louis Leroy 28 Sep '09

28 Sep '09

I'm trying the make_map example from the doc, nearly verbatim : #include <boost/fusion/container/generation/make_map.hpp> #include <boost/fusion/include/make_map.hpp> int main() { using namespace boost::fusion; make_map( make_pair<int>('X') , make_pair<double>("Men")); } Both msvc9 and g++ 4.4.1 tell me that there is no make_map function that takes two parameters. g++ mentions that it sees a parameter-less make_map(). I use Boost 1.40. What's up ? Thanks, J-L

3 2

[fusion] "modify" value in map
by Jean-Louis Leroy 27 Sep '09

27 Sep '09

How can I create a map that has the same content as another rmap, except for a given key which has a new associated value ? I'm looking for something like replace_key<Key>(Sequence, Value) in the doc but I cannot find it. I think I can build it with push_back(erase_key<Key>(seq), make_pair<Key>(newval)) but if it already exists... Thanks, J-L

2 2

Regex problem: cannot parse terms containing OR
by Ramon F Herrera 27 Sep '09

27 Sep '09

I have been able to parse stuff more complicated than this, but now I am stuck with something seemingly simpler. The expression being parsed is the common sequence: 1,2-5,7,8-11 Question 1: Notice my approach. I first match the whole expression, with "regex_match", to make sure that it is valid (that works great). Next, I use "regex_iterator" to break down the parts. Is that good practice? Am I being inefficient/redundant? Question 2: My code below only extracts "range terms" ("x-y"), for some reason I cannot extract "number terms". As a workaround, I can always feed my data like this: 1-1,2-5,7-7,8-11 but, after a lot of tries, would love to learn how to do this properly. TIA, -Ramon ----------------------------------------------------- #include <iostream> #include <boost/regex.hpp> using namespace std; bool term_callback(const boost::match_results<std::string::const_iterator>& what) { for (unsigned int i = 0; i < what.size(); i++) { cout << "what[" << i << "]: " << what[i].str() << endl; cout << "---------" << endl; } return true; } int main(int argc, char *argv[]) { const char hyphen = '-'; const char left_paren = '('; const char right_paren = ')'; const char bar = '|'; const char comma = ','; const char star = '*'; const string number = "[0-9]+"; const string range = number + hyphen + number; const string term = left_paren + number + bar + range + right_paren; const string sequence = term + bar + left_paren + term + comma + right_paren + star + term; boost::regex expression(sequence); boost::regex piece(range); boost::cmatch matches; char argument[1024]; strcpy(argument, argv[1]); if (!boost::regex_match(argument, matches, expression)) { cerr << "There is no match" << endl; return 1; } string text = argument; boost::sregex_iterator m1(text.begin(), text.end(), piece); boost::sregex_iterator m2; for_each(m1, m2, &term_callback); return 0; }

3 2

Re: [Boost-users] [Test] How to treat memory leaks as errors?
by Василий Старостин 27 Sep '09

27 Sep '09

Hi! >> However, Boost.Test doesn't return non-zero error code when leaks >> detected (as it does this when tests fail). >> How can I make my testing program to return non-zero error code if memory leaked? >Because Boost.Test does not contain anything that can test for memory >leaks. You might want to use something like Google's tcmalloc and so >forth as it can tell you about all that, and you can just add a >Boost.Test testcase at the end of the tests to check for memory leaks >using it. Boost.Test does contain something to test memory leaks since I get such warnings when I loose memory in my Boost.Test-based program :) I know this is _CRT functions (I use MSVC9). And the function, which Boost.Test uses to inform about leaks under Windows, SIGNALS that we have m-leaks in a program. Moreover, Boost.Test can control this function using (to use or not to use). So I expected that Boost.Test can use this function return code and increase errors counter for leaks. If it cannot, this can be improvement for the future. WBR, Vasily P.S. _CrtDumpMemoryLeaks?

3 6

Re: [Boost-users] Sockets and structures?
by Brian Wood 27 Sep '09

27 Sep '09

Daniel F. Savarese writes: > To the OP, I would advise you not take statements such as the above as > gospel without performing measurements yourself. My experience[1] is > directly contrary to the above. I find boost::serialization execution > time to be negligible (for binary archives) and entirely suitable for high > performance networking applications even if that was not the creator's > original objective (a credit to Robert Ramey's skill). With respect for the library and Ramey, I'd like to point out that results obtained with the Boost Serialization library are frequently less efficient than those from the C++ Middleware Writer -- http://webEbenezer.net/comparison.html<http://webebenezer.net/comparison.html> . Additionally, the C++ Middleware Writer automates the job of writing serialization functions for structs/classes unless it is directed not to do so. Regards, Brian Wood Ebenezer Enterprises www.webEbenezer.net "Then Samuel took a rock and set it up between Mizpah and Shen. He named it Ebenezer [Rock of Help] and said, 'Until now the L-RD has helped us.'"

2 1

[Test] Compiler error/missing entry point?
by Michael Spece Ibanez 26 Sep '09

26 Sep '09

I installed the Boost.Test library binary using bootstrap.sh. Is the following a valid program? If not, what is wrong? Is there a missing entry point and, if so, how do I run the test? Source: #define BOOST_TEST_MODULE ExampleTestModule #include <boost/test/unit_test.hpp> BOOST_AUTO_TEST_CASE( my_test ) { BOOST_CHECK(1); } Error: /cygdrive/c/DOCUME~1/M/LOCALS~1/Temp/ccIao0DX.o:TestExample.cc:(.text+0x155): undefined reference to `boost::unit_test::framework::master_test_suite()' /cygdrive/c/DOCUME~1/M/LOCALS~1/Temp/ccIao0DX.o:TestExample.cc:(.text+0x232): undefined reference to `boost::unit_test::unit_test_log_t::set_checkpoint(boost::unit_test::basic_cstring<char const>, unsigned int, boost::unit_test::basic_cstring<char const>)' /cygdrive/c/DOCUME~1/M/LOCALS~1/Temp/ccIao0DX.o:TestExample.cc:(.text+0x317): undefined reference to `boost::test_tools::tt_detail::check_impl(boost::test_tools::predicate_result const&, boost::unit_test::lazy_ostream const&, boost::unit_test::basic_cstring<char const>, unsigned int, boost::test_tools::tt_detail::tool_level, boost::test_tools::tt_detail::check_type, unsigned int, ...)' /cygdrive/c/DOCUME~1/M/LOCALS~1/Temp/ccIao0DX.o:TestExample.cc:(.text+0x4e0): undefined reference to `boost::unit_test::ut_detail::auto_test_unit_registrar::auto_test_unit_registrar(boost::unit_test::test_case*, unsigned long)' /cygdrive/c/DOCUME~1/M/LOCALS~1/Temp/ccIao0DX.o:TestExample.cc:(.text$_ZN5boost9unit_test14make_test_caseERKNS0_9callback0INS0_9ut_detail6unusedEEENS0_13basic_cstringIKcEE[boost::unit_test::make_test_case(boost::unit_test::callback0<boost::unit_test::ut_detail::unused> const&, boost::unit_test::basic_cstring<char const>)]+0x54): undefined reference to `boost::unit_test::ut_detail::normalize_test_case_name(boost::unit_test::basic_cstring<char const>)' /cygdrive/c/DOCUME~1/M/LOCALS~1/Temp/ccIao0DX.o:TestExample.cc:(.text$_ZN5boost9unit_test14make_test_caseERKNS0_9callback0INS0_9ut_detail6unusedEEENS0_13basic_cstringIKcEE[boost::unit_test::make_test_case(boost::unit_test::callback0<boost::unit_test::ut_detail::unused> const&, boost::unit_test::basic_cstring<char const>)]+0xa6): undefined reference to `boost::unit_test::test_case::test_case(boost::unit_test::basic_cstring<char const>, boost::unit_test::callback0<boost::unit_test::ut_detail::unused> const&)' /cygdrive/c/DOCUME~1/M/LOCALS~1/Temp/ccIao0DX.o:TestExample.cc:(.text$_ZN5boost9unit_test15unit_test_log_tD1Ev[boost::unit_test::unit_test_log_t::~unit_test_log_t()]+0xb): undefined reference to `vtable for boost::unit_test::unit_test_log_t' /cygdrive/c/DOCUME~1/M/LOCALS~1/Temp/ccIao0DX.o:TestExample.cc:(.text$_ZN5boost9unit_test15unit_test_log_tC1Ev[boost::unit_test::unit_test_log_t::unit_test_log_t()]+0x21): undefined reference to `vtable for boost::unit_test::unit_test_log_t' /usr/lib/gcc/i686-pc-cygwin/3.4.4/../../../libcygwin.a(libcmain.o):(.text+0xab): undefined reference to `_WinMain@16' collect2: ld returned 1 exit status

2 5

[Test] catch_system_errors default value ...
by Georg Sauthoff 26 Sep '09

26 Sep '09

Hi, the default of 'catch_system_errors = yes' when using the boost unit testing framework is a bit problematic. Apparently, it installs a signal handler for SIGCHILD. I didn't notice this under Linux. But under Solaris I've got an interrupted waitpid system call in one test case[1]. Well, and the trace looks really funny, if waitpid is interrupted by ... SIGCHILD! (the testcase basically does a fork/exec/waitpid) Thus, the default of this parameter is at least a bit counter-intuitive. Perhaps, you[2] consider to change this default in the future. Best regards Georg Sauthoff [1] I assume, that utf does this on Linux in such a way, that waitpid is automatically resumed [2] I mean the utf maintainer, using gmane, I am not able to CC ...

2 1