September 2009 - Boost-users

[Intrusive] List Manipulation and Management
by Freddie Witherden 26 Sep '09

26 Sep '09

Hi all, I am moderningising/porting some old Pascal code. The code makes extensive use of an intrusive singly-linked structure termed a node. In order to replicate this idiom in C++ I have been investigating the Intrusive library. However, I have several questions about list manipulation and memory management. Firstly, in the code being ported all nodes can be thought of as being heap allocated. The following is extremely common (pseudo-code): Node* afunction(...); [...] Node* tail; link(tail, afunction(...)); This is rather nice as in O(1) you can append a returned list to your current list. However, I am unsure how I would accomplish this in Boost Intrusive. I can not seem to find a way to create the somewhat fragile intrusive::list from just a node (be it singly or doubly linked). The list class also seems to take T& as opposed to T* when pusing to the front/back of the list. My experience of intrusive structure in C was that you'd always take a pointer when adding a new link. Secondly, when dealing with heap allocated items, which need to be deleted, the current code uses a while loop to call each nodes destructor. What is the intrusive equivalent? Of course, I am open to ideas/suggestions about better ways to accomplish the above tasks :) Regards, Freddie.

2 4

[random]: several issues
by Thomas Mang 26 Sep '09

26 Sep '09

HI, After some time I came back to the boost random library, and several things I noticed earlier have not changed since years ago. I hereby ask what the present state of ideas regarding these ideas is. Here's my list: a) The library itself provides files for random deviates of distributions not given in the documentation (e.g. poisson, gamma) etc. I find it truly sad that implementation and documentation is that out of synchronisation b) Several distributions require the engine to return a uniform deviate between [0,1), while other's don't have this prerequisite. I find this extremely error prone and purely documented (it is documented, but it should be so clearer, and especially, louder). Worse, I find it even harder (or impossible) to find out which range the engines return. I am not sure if there is any engine returning this range per se, or if I have always to go through uniform_01. Worst of all, there is neither a compile time check nor a runtime check if that [0,1) result-requirement of the engine holds (as far as I have seen the code) - I absolutely fail to see why such a critical assert is completely missing, also given the poor state of the documentation. The poisson and gamma for example also fall in this category, but are not documented at all. I consider this highly dangerous. c) Random numbers are tightly linked to statistical distributions, offered by the library of it's own. Wouldn't it be convenient to try to integrate the whole distribution part of the random numbers more closely into that library? Presently they are too confusingly standalone. d) Is there anyone actually responsible for the random library at the moment? Is it still under active development / maintained, or is old code simply propagating from release to release ? [No offense intended, but above issues let me have doubts regarding active management). thanks, Thomas

4 6

[Test] Test suite timeout.
by Dmitry V. Krivenok 26 Sep '09

26 Sep '09

Hello! I have a question about timeouts implementation in Boost::Test. Please take a look at a simple example I wrote to demonstrate the problem: krivenok@develop2 16:50:52 /tmp/boost_test $ cat main.cpp #include <boost/test/unit_test.hpp> using namespace boost::unit_test; using boost::unit_test::framework::master_test_suite; //////////////////////////////////////////////////////////////////////////////// void some_test_case() { BOOST_TEST_CHECKPOINT("About to enter an infinite loop!"); while(1); } ////////////////////////////////////////////////////////////////////////////// void another_test_case() { BOOST_TEST_CHECKPOINT("About to enter an infinite loop 2!"); while(1); } ////////////////////////////////////////////////////////////////////////////// boost::unit_test::test_suite* create_test_suite() { boost::unit_test::test_suite *ts = new boost::unit_test::test_suite("Some test suite"); if(!ts) return 0; ts->add(BOOST_TEST_CASE(&some_test_case), 0, 5); ts->add(BOOST_TEST_CASE(&another_test_case), 0, 10); return ts; } /////////////////////////////////////////////////////////////////////////////// test_suite* TestSuitesInit(int argc, char* argv[]) { master_test_suite().add(create_test_suite(), 0, 3); return 0; } /////////////////////////////////////////////////////////////////////////////// int main(int argc, char** argv) { return unit_test_main(&TestSuitesInit, argc, argv); } /////////////////////////////////////////////////////////////////////////////// krivenok@develop2 16:50:53 /tmp/boost_test $ I compiled it as follows: krivenok@develop2 16:56:12 /tmp/boost_test $ g++-4.4.1 -Wall -I/usr/local/dev/boost-1.40.0/include main.cpp /usr/local/dev/boost-1.40.0/lib/libboost_unit_test_framework-mt.a krivenok@develop2 16:56:15 /tmp/boost_test $ And the run: krivenok@develop2 16:58:17 /tmp/boost_test $ time ./a.out Running 2 test cases... unknown location(0): fatal error in "some_test_case": signal: SIGALRM (timeout while executing function) main.cpp(10): last checkpoint: About to enter an infinite loop! unknown location(0): fatal error in "another_test_case": signal: SIGALRM (timeout while executing function) main.cpp(18): last checkpoint: About to enter an infinite loop 2! *** 2 failures detected in test suite "Master Test Suite" real 0m15.002s user 0m14.920s sys 0m0.030s krivenok@develop2 16:58:36 /tmp/boost_test $ According to timings measurement I concluded that Boost::Test supports timeouts on per-test-case basis. Test suite timeout (3 sec) is ignored. Is it a bug? P.S. I also found that timeouts are implemented using alarm system call: krivenok@develop2 16:58:36 /tmp/boost_test $ strace -fF ./a.out 2>&1 | grep alarm alarm(0) = 0 alarm(5) = 0 alarm(0) = 0 alarm(0) = 0 alarm(10) = 0 alarm(0) = 0 krivenok@develop2 17:02:03 /tmp/boost_test $ As I know you can't start two or more timers using alarm since alarm uses exactly one internal timer. The following simple program demonstrates the problem: krivenok@develop2 17:06:38 /tmp/boost_test/alarm $ cat main.c #include <unistd.h> int main() { alarm(2); alarm(5); while(1); return 0; } krivenok@develop2 17:06:42 /tmp/boost_test/alarm $ gcc-4.4.1 main.c krivenok@develop2 17:06:44 /tmp/boost_test/alarm $ time ./a.out Alarm clock real 0m5.001s user 0m4.990s sys 0m0.000s krivenok@develop2 17:06:51 /tmp/boost_test/alarm $ strace -fF ./a.out 2>&1 | grep alarm alarm(2) = 0 alarm(5) = 2 krivenok@develop2 17:07:00 /tmp/boost_test/alarm $ This makes the implementation of mixed timeouts difficult and unreliable. For example, consider the following test tree: MasterTestSuite (timeout = T0) -- TestSuite1 (timeout = T1) -- TestCase11(timeout = T11) -- TestCase12(timeout = T12) -- TestSuite2 (timeout = T2) -- TestCase21(timeout = T21) -- TestCase212(timeout = T22) Actual timeout of any test case depends on timeouts of all nodes in the path. -- Sincerely yours, Dmitry V. Krivenok Orange System Co., Ltd. Saint-Petersburg, Russia work phone: +7 812 332-32-40 cellular phone: +7 921 576-70-91 e-mail: krivenok(a)orangesystem.ru web: http://www.orangesystem.ru skype: krivenok_dmitry jabber: krivenok_dmitry(a)jabber.ru icq: 242-526-443

2 1

unit test: BOOST_FIXTURE_TEST_CASE_TEMPLATE
by Hicham Mouline 26 Sep '09

26 Sep '09

Hello, I would like to request for the addition of this macro to the boost unit tests library. I have a mpl::vector_c<....> and I would like to run a test case for the type being each of the mpl integrals. and test case instantiations would have the same fixture. I don't see a workaround but for now I'll just use a runtime loop. Rds,

2 1

fork in boost?
by pedro chaparro 26 Sep '09

26 Sep '09

hi, there is some function in boost libraries equivalent to "fork" in linux? thanks -- Ing Pedro Alonso Chaparro Valero Ciudad Politecnica de la Innovación iTEAM - Mobile Communications Group Polytechnic University of Valencia C\ Camino de Vera S/N, Edificio 8G 46022 Valencia, Spain

3 2

Argument error after upgrading to Snow Leopard
by s seelenluft 26 Sep '09

26 Sep '09

Hi, after upgrading to Snow Leopard I get following error message compiling even a simple Hello World program with bjam: *** argument error * rule SHELL ( command : * ) * called with: ( ) * missing argument command (builtin):see definition of rule 'SHELL' being called ... The .cpp file is: #include <iostream> int main() { cout "Hello World!" << endl; return 0; } The Jamroot file contains: exe HelloWorld : HelloWorld.cpp ; I have: - bjam installed in /usr/local/bin - boost-build libraries in /usr/share/boost-build - in /usr/share/boost-build/user-config.jam the line: using darwin; This is as far as I know exactly the same as I had it under Leopard. What am I missing here? Thanks. Markus

3 5

[python] Passing pointers as arguments without copying.
by Germán Diago 26 Sep '09

26 Sep '09

Hello. The question is simple, but reading the documentation I cannot understand how to do this: class MyClass : boost::noncopyable { void f(MyClass2 * c2); }; MyClass2 is not copyable either. So I want to be able to wrap that function, but avoid the copying of MyClass2 when called from python. I want to pass the c++ object directly, not a copy of it. I'm not sure if this can be done with ptr() or some kind of call policy, but I don't understand the documentation 100%. Any help will be appreciated. Thanks.

3 3

Week of the Year calculations?
by Ramon F Herrera 26 Sep '09

26 Sep '09

I looked at the Date_Time library but couldn't find this. Need to convert a date into the corresponding week of the year and vice versa. I realize that such thing varies and plan to use the standard most common in the USA. TIA, -Ramon

1 1

[exception] nested exceptions: lost information?
by Pascal Z. 26 Sep '09

26 Sep '09

Hello I'm trying to nest exceptions using boost.Exception, but it seems like some information (in this case the message returned via what() ) is lost on the way. #include <iostream> #include <boost/exception.hpp> typedef boost::error_info<struct tag_nested_exception, boost::exception_ptr> nested_exception; struct file_open_error : public boost::exception, public std::exception {}; int open_file(const std::string& filename) { try { // ... throw std::runtime_error("Some internal error."); } catch(const std::exception& error) { throw file_open_error() << nested_exception(boost::copy_exception(error)); } } int main(int argc, char* argv[]) { try { int file = open_file("foo.txt"); file++; } catch(const boost::exception& error) { try { boost::rethrow_exception(*boost::get_error_info<nested_exception>(error)); } catch(const std::exception& nested) { std::cout << "the nested error is: " << nested.what() << std::endl; } } return 0; } I except the output to be "the nested error is: Some internal error." but it outputs "the nested error is: std::exception". What can I do to make it work? Greetings Pascal

2 2

std::accumulate with condition?
by Archie14 26 Sep '09

26 Sep '09

If a given class defines operator + and there is a vector of instances of that class: how I can use std::accumulate to count sum against this vector for the elements that satisfy certain conditions? class A { public: A(useinsum) : _useinsum(useinsum) {}; bool useinsum() {return _useinsum;} int operator + (int a) {return _i + a;} private bool _useinsum; int _i; } std::vector<A> lst; will that work? std::accumulate(lst.begin(), lst.end(), 0, boost::lambda::bind(&A::useinsum, boost::lambda::_1) == false)

2 1