- Boost-users - lists.boost.org

configuring a library to work with multiple versions of boost
by Ross Boylan 01 Nov '05

01 Nov '05

I have a library that uses boost and also uses a pre-release version of the ptr_container (header only) library that eventually made it into boost 1.33 (I think). I would like people to be able to compile my code with various versions of boost. I've had it working with 1.31 and 1.32, and expect to have it working with 1.33 when that reaches my distribution. I have a subdirectory with the pre-release ptr_container library and some supporting code from elsewhere in boost. This is in a "boost" subdirectory of my project, which has included the directory above that unconditionally in my compilations. With 1.33 I think I want to ignore my local stuff completely. I'm using autoconf and libtool. Are there autoconf macros available for detecting the boost libraries? Do people have any suggestions about the best way to proceed? One possibility that occurs to me is to test for the version of the library at configure time, and make the include path depend on the results. Thanks. Ross Boylan

2 2

ptr_map
by Oleg Smolsky 01 Nov '05

01 Nov '05

Hello Thorsten, thanks for the ptr_vector clarifications. I have another question about ptr_map: I can declare a map iterator in the following way: ptr_map<string, SomeClass>::iterator i = m.begin(); Now, I'd expect to be able to compile the following, as with a normal map: string s = i->first; i->second.SomeMethod(); ...assuming SomeClass::SomeMethod() exists and is declared as public. However, "first" and "second" don't seem to exist in ptr_map. Could you shed some light on how I can iterate your map please? Best regards, Oleg.

2 3

[test]
by Paul Giaccone 01 Nov '05

01 Nov '05

Is the use of BOOST_CHECK_EQUAL not permitted in init_unit_test_suite(int argc, char* argv[])? The following code throws an exception and reports the error "Boost.Test internal framework error: unknown reason" boost::unit_test::test_suite* init_unit_test_suite(int argc, char* argv[]) { BOOST_CHECK_EQUAL(foo(argc, argv), 0); boost::unit_test::test_suite* test = BOOST_TEST_SUITE("My test"); test -> add(BOOST_TEST_CASE(&test1)); test -> add(BOOST_TEST_CASE(&test2)); return test; } when the following works without any problems: boost::unit_test::test_suite* init_unit_test_suite(int argc, char* argv[]) { int result = foo(argc, argv); assert(result == 0); boost::unit_test::test_suite* test = BOOST_TEST_SUITE("My test"); test -> add(BOOST_TEST_CASE(&test1)); test -> add(BOOST_TEST_CASE(&test2)); return test; } I have no problem with using the latter code, but I am surprised that I get an exception in the former code and that it is caught by "..." and not by the library. If BOOST_CHECK_EQUAL (and BOOST_CHECK, BOOST_CHECK_CLOSE, etc) is not allowed in init_unit_test_suite(), is this fact documented? Should it generate an exception that can be caught by the library and reported with a useful message? Thanks, Paul

2 2

Re: [Boost-users] [Multi-Index] composite_key problems
by JOAQUIN LOPEZ MU?Z 01 Nov '05

01 Nov '05

----- Mensaje original ----- De: Andreas Volz <lists(a)brachttal.net> Fecha: Martes, Noviembre 1, 2005 0:28 am Asunto: Re: [Boost-users] [Multi-Index] composite_key problems > I got it working! Perhaps I've some more questions, but for now I > couldcontinue my work. Excellent! Good luck with your project. > > Did someone create a SQL language wrapper around multi_index? I think > this could be a very great feature. People that know SQL could really > easy use it without learning too much other stuff. There's currently a very active debate going on with respect to the merits of two libraries which offer a more SQL-friendly interface to handle multi-index data structures (none of them is actually SQL compatible, though.) Google for RML site:boost.org to get some links to the discussion. RML site is http://visula.org/relational/ while RTL stuff can be found at http://boost-consulting.com/vault/index.php? &direction=0&order=&directory=RTL > I thought about > using a lightweight SQL lib (e.g. sqlite) before I found multi_index. > But I don't save my data into a file, so multi_index fits perfect. > You can also save/load multi_index_containers contents to/from a file via Boost.Serialization. Check http://boost.org/libs/multi_index/doc/advanced_topics.html#serializatio n Joaquín M López Muñoz Telefónica, Investigación y Desarrollo

2 1

Visual C++ 2005 (8.0) support
by Ovanes Markarian 01 Nov '05

01 Nov '05

Hi all, did someone test boost 1.33 with Visual C++ 2005(8.0)? I am wondering how I can compile the library using jam. Should I use VC++ 7.1 option? Another question. I derived an iterator from the iterator_facade template. As I have recently found out VC++ 8.0 did not support friend template classes, e.g. template<class ....> class my_type { template <class> friend class my_type; ... }; This will not work in VC++ 8.0. Did you know about this issue? Should this be filed as a bug at MS? With Kind Regards, Ovanes Markarian

5 5

Re: [Boost-users] problem with multi_index_container -> find
by JOAQUIN LOPEZ MU?Z 01 Nov '05

01 Nov '05

Hi Oliver, ----- Mensaje original ----- De: Oliver.Kowalke(a)infineon.com Fecha: Martes, Noviembre 1, 2005 10:38 am Asunto: [Boost-users] problem with multi_index_container -> find > Hello, > I don't understand why this code doesn't compile (gcc-3.3.1) '... > id_index.find( id) ...' ? > Regards, > Oliver [...] > phonebook::iterator itr = id_index.find( id); // <<== doesn't id_index.find() returns a phonebook::nth_index_iterator<1>::type. phonebook::iterator is shorthand for phonebook::nth_index_iterator<0>::type, a different type. The following should work phonebook::nth_index_iterator<1>::type itr = id_index.find(id); or alternatively phonebook::nth_index<1>::type::iterator itr = id_index.find(id); Hope this helps, Joaquín M López Muñoz Telefónica, Investigación y Desarrollo

1 0

How to build Boost for Cygwin (asked the noob)
by Olumide 01 Nov '05

01 Nov '05

Hi - Would someone please give me some (SIMPLE) pointers on how to build Boost for Cygwin. Thanks, - Olumide

2 1

[threads] read_write_mutex and condtion
by Phil Endecott 01 Nov '05

01 Nov '05

Dear All, Is it possible to use a boost::condition in combination with a read_write_mutex? I've tried but failed. This is with 1.32.0. My test code follows. As you can see it can use either a mutex or a read_write_mutex depending on which bits are commented-out. It works as expected with a mutex, but with a read_write_mutex I get a compiler error, shown below. Perhaps I am missing some fundamental reason why conditions can't work with read_write_mutexes. Can someone explain? Thanks, Phil. #include <string> #include <iostream> #include <boost/thread/mutex.hpp> #include <boost/thread/read_write_mutex.hpp> #include <boost/thread/condition.hpp> #include <boost/thread/thread.hpp> #include <boost/bind.hpp> #include <unistd.h> using namespace std; using namespace boost; struct shared_thing { string data; //typedef mutex mx_t; //typedef mutex::scoped_lock read_lock; //typedef mutex::scoped_lock write_lock; typedef read_write_mutex mx_t; typedef read_write_mutex::scoped_read_lock read_lock; typedef read_write_mutex::scoped_write_lock write_lock; mx_t mx; condition cn; shared_thing::shared_thing(void): //mx() mx(read_write_scheduling_policy::writer_priority) {} void writer(void) { while(1) { sleep(1); { write_lock l(mx); data += "."; cn.notify_all(); } } } void reader(void) { while(1) { read_lock l(mx); cn.wait(l); cout << data << endl; } } }; int main(int argc, char* argv[]) { shared_thing t; thread wr_thread(bind(&shared_thing::writer,&t)); t.reader(); } $ g++ -pthread -O -Wall -o condvar condvar.cc -lboost_thread /usr/include/boost/thread/detail/read_write_lock.hpp: In member function `void boost::condition::wait(L&) [with L = boost::detail::thread::scoped_read_lock<boost::read_write_mutex>]': condvar.cc:51: instantiated from here /usr/include/boost/thread/detail/read_write_lock.hpp:297: error: ` boost::read_write_mutex&boost::detail::thread::scoped_read_lock<boost::read_write_mutex>::m_mutex ' is private /usr/include/boost/thread/condition.hpp:92: error: within this context

3 2

problem with multi_index_container -> find
by Oliver.Kowalke＠infineon.com 01 Nov '05

01 Nov '05

Hello, I don't understand why this code doesn't compile (gcc-3.3.1) '... id_index.find( id) ...' ? Regards, Oliver #include <iostream> #include <string> #include <utility> #include <boost/multi_index_container.hpp> #include <boost/multi_index/key_extractors.hpp> #include <boost/multi_index/ordered_index.hpp> #include <boost/tuple/tuple.hpp> namespace mi = boost::multi_index; struct phonebook_entry { int id; std::string family_name; std::string given_name; phonebook_entry( int id_, std::string const& family_name_, std::string const& given_name_) : id( id_), family_name( family_name_), given_name( given_name_) {} }; typedef mi::multi_index_container< phonebook_entry, mi::indexed_by< mi::ordered_non_unique< mi::composite_key< phonebook_entry, mi::member< phonebook_entry, std::string, & phonebook_entry::family_name >, mi::member< phonebook_entry, std::string, & phonebook_entry::given_name > > >, mi::ordered_unique< mi::member< phonebook_entry, int, & phonebook_entry::id > > > > phonebook; void test_multi_index() { phonebook pb; phonebook::nth_index< 0 >::type & name_index = pb.get< 0 >(); std::string family_name("Mann"); std::string given_name("Thomas"); std::pair< phonebook::iterator, phonebook::iterator > p = name_index.equal_range( boost::make_tuple( family_name, given_name) ); while ( p.first != p.second) { std::cout << p.first->family_name << ", " << p.first->given_name << std::endl; ++p.first; } phonebook::nth_index< 1 >::type & id_index = pb.get< 1 >(); int id = 1; phonebook::iterator itr = id_index.find( id); // <<== doesn't // error: conversion from ` boost::multi_index::detail::index_iterator<boost::multi_index::detail::o rdered_index_node<boost::multi_index::detail::index_node_base<phonebook_ entry> > >' to non-scalar type ` // boost::multi_index::detail::index_iterator<boost::multi_index::detail::o rdered_index_node<boost::multi_index::detail::ordered_index_node<boost:: multi_index::detail::index_node_base<phonebook_entry> > > >' //requested phonebook::iterator end = id_index.end(); if ( itr != end) std::cout << itr->family_name << ", " << itr->given_name << std::endl; }

1 0

Re: [Boost-users] [Multi-Index] composite_key problems
by JOAQUIN LOPEZ MU?Z 31 Oct '05

31 Oct '05

----- Mensaje original ----- De: Andreas Volz <lists(a)brachttal.net> Fecha: Lunes, Octubre 31, 2005 7:27 pm Asunto: Re: [Boost-users] [Multi-Index] composite_key problems > The questions seems to by stupic, but how > could I > access the data looked up by B alone? What should I assign to the > left? > > ??? = es.get<1>().find(b); > > An iterator or what else? This returns an iterator, yep. Please notice that es.get<1>() returns a reference to an index, which is, in many respects, entirely similar to a regular std::set. So you have find(), equal_range() etc. just like in the STL case, with equivalent syntax and semantics. You might want to give the Boost.MultiIndex tutorial a further read --anyway, don't hesitate to come back here if things aren't still clearer. Joaquin M Lopez Muñoz Telefonica, Investigacion y Desarrollo

2 1