June 2008 - Boost-users - lists.preview.boost.org

Re: [Boost-users] Quick questions on string algorithms
by Yang Zhang 12 Jun '08

12 Jun '08

*bump*, and also.... When using the split_iterator, is there a way to find a match's offset into the string? (It seems I can only dereference the iterator to get the string, when I would instead like to find the position.) On Fri, Jun 6, 2008 at 4:02 PM, Yang Zhang <yanghatespam(a)gmail.com> wrote: > > Is there a way to split a string only n times? (a la Java's String.split(String, int)) Can one do this in reverse? Is it possible to do this using the split iterator? (Simple example is if I'd like to implement Unix basename and dirname using these algos.) > > Is there anything that could interpret a (C-like) string literal? E.g.: > > "\"hello,\\nworld!\"" -> "hello,\nworld!" > > Thanks in advance! > -- > Yang Zhang > http://www.mit.edu/~y_z/ > -- Yang Zhang http://www.mit.edu/~y_z/

1 0

intrusive avl_set and offset_ptr
by Lothar Werzinger 12 Jun '08

12 Jun '08

Hi, I try to use intrusive avl_set to put the set into a memory mapped file (I manage the file myself using just mapped_region. If I declare my node without offset_ptr my test program works as expected except that the pointers in the file are absolute. If I use offset_ptr I get a runtime assertion (see below). What I don't understand is why the assertion complains about uniqueness. Any help is appreciated! struct BalanceNode : public api::helper::BalanceKey // ,public bin::avl_set_base_hook<bin::optimize_size<true> > ,public bin::avl_set_base_hook< bin::optimize_size<true>, bin::void_pointer< bip::offset_ptr<void> > > { api::values::Decimal m_amount; }; // Define an avl_set using the base hook typedef bin::avl_set< BalanceNode > BalanceSet; /opt2/linux/x86_64/include/boost-1_35/boost/intrusive/avltree.hpp:618: boost::intrusive::tree_iterator<boost::intrusive::avltree_impl<Config>, false> boost::intrusive::avltree_impl<Config>::insert_unique_commit(typename std::iterator_traits<typename boost::intrusive::detail::eval_if_c<boost::intrusive::avltree_impl<Config>::external_value_traits, boost::intrusive::detail::eval_value_traits<typename Config::value_traits>, boost::intrusive::detail::identity<typename Config::value_traits> >::type::pointer>::reference, const typename boost::intrusive::avltree_algorithms<typename boost::intrusive::detail::eval_if_c<boost::intrusive::avltree_impl<Config>::external_value_traits, boost::intrusive::detail::eval_value_traits<typename Config::value_traits>, boost::intrusive::detail::identity<typename Config::value_traits> >::type::node_traits>::insert_commit_data&) [with Config = boost::intrusive::avl_setopt<boost::intrusive::detail::base_hook_traits<tradescape::balance::BalanceNode, boost::intrusive::avltree_node_traits<boost::interprocess::offset_ptr<void>, true>, safe_link, boost::intrusive::default_tag, 6>, std::less<tradescape::balance::BalanceNode>, long unsigned int, true>]: Assertion `node_algorithms::unique(to_insert)' failed. Lothar -- Lothar Werzinger Dipl.-Ing. Univ. framework & platform architect Tradescape Inc. - Enabling Efficient Digital Marketplaces 1754 Technology Drive, Suite 128 San Jose, CA 95110 web: http://www.tradescape.biz

2 4

Re: [Boost-users] [boost-users][multi-index] iterating though index keys
by JOAQUIN M. LOPEZ MUÑOZ 12 Jun '08

12 Jun '08

_______________________________________ De: boost-users-bounces(a)lists.boost.org [boost-users-bounces(a)lists.boost.org] En nombre de Igor R [boost.lists(a)gmail.com] Enviado el: jueves, 12 de junio de 2008 19:58 Para: boost-users(a)lists.boost.org Asunto: Re: [Boost-users] [boost-users][multi-index] iterating though indexkeys > Hi Joaquín, > > > for(Entries::iterator > > it_key=entries.begin(),end=entries.end();it_key!=end;){ > > // get all elements with key it_key->someForeignKey > > std::pair<Entries::iterator,Entries::iterator> values= > > entries.equal_range(it_key->someForeignKey); > > while(values.first!=values.second){ > > // process element > > ++values.first; > > } > > // next key is just that of the element right after the current value > > group > > it_key=values.second; > > } > > Thank you! It seems to be exactly what I need. > However, I'm not sure I got the principle right: how come that "next > key is just that of the element right after the current value"? Is > this the general behavior of equal_range - despite of the kind of the > index (ordered/hashed, unique/non-unique)? Yep, and the behavior of STL containers like std::multiset, for that matter. This is easily seen if you consider that indices of a multi_index_container (and STL containers) guarantee that elements with equivalent key are grouped together. So, if you hop from a group of same-key elements to the next you end up visiting all the keys present in the container. Joaquín M López Muñoz Telefónica, Investigación y Desarrollo

1 0

Re: [Boost-users] [boost-users][multi-index] iterating though index keys
by Igor R 12 Jun '08

12 Jun '08

Hi Joaquín, > for(Entries::iterator > it_key=entries.begin(),end=entries.end();it_key!=end;){ > // get all elements with key it_key->someForeignKey > std::pair<Entries::iterator,Entries::iterator> values= > entries.equal_range(it_key->someForeignKey); > while(values.first!=values.second){ > // process element > ++values.first; > } > // next key is just that of the element right after the current value > group > it_key=values.second; > } Thank you! It seems to be exactly what I need. However, I'm not sure I got the principle right: how come that "next key is just that of the element right after the current value"? Is this the general behavior of equal_range - despite of the kind of the index (ordered/hashed, unique/non-unique)?

1 0

[boost-users][multi-index] iterating though index keys
by Igor R 12 Jun '08

12 Jun '08

Hi, I'm trying to replace few data structures and types defined in my application in terms of std containers, with the mutli_index_container. Some of those types look like this: struct Entry { int someForeignKey; int data1; std::string data2; //.... }; typedef std::deque<Entry> Entries; typedef std::map<int, Entries> EntriesByForeignKey; I'd like to rewrite the above map as follows: typedef mi::multi_index_container < Entry, mi::indexed_by < mi::hashed_non_unique<mi::tag<foreignKey>, mi::member<Entry, int, &Entry::someForeignKey> > > > Entries; typedef Entries::index<foreignKey>::type EntriesByForeignKey; Now, I'd like to iterate through the *keys* of this index, and for each key - iterate through all its values. IIUC, the second part can be accomplished like this: { Entries entries; EntriesByForeignKey::iterator begin = entries.get<foreignKey>().find(theKey), end = entries.get<foreignKey>().end(); for_each (begin, end, ....); } ...but how can I iterate through all the keys of this index? Or should I define another index for this purpose (sequenced one?) ? Thank you, Igor'.

2 1

phoenix question
by Hansi 12 Jun '08

12 Jun '08

Hello, at the moment I'm reading the documentation from Phoenix. Now I have read in the first pages that there are 2 possibilities to define a functor 1) find_if(c.begin(), c.end(), arg1 % 2 == 1) 2) struct is_odd_ { template <typename ArgT> struct result { typedef bool type; }; template <typename ArgT> bool operator()(ArgT arg1) const { return arg1 % 2 == 1; } }; function<is_odd_> is_odd; Okay this 2 possibilities I understand. But There is also written that for much more complex functions the second possibility is more efficient. Can anyone say why this is more efficient? Best regards Hansjörg

1 0

customized circular_buffer (thread safe implementation)
by Selçuk Giray Özdamar 12 Jun '08

12 Jun '08

Hi everyone, I wrote my template based CircularBuffer class. Here the primary aim is to make everything thread safe. Multiple concurrent reads are possible, but only exclusive write is possible. Here we have index operator overloaded. As usual index operator has 2 versions, const and non-const versions. T& operator [] (unsigned index); // return as reference T operator [] (unsigned index) const; // return by value In contrast, standart library container classes return by const reference in const methods. Here I'm returning by value. Below you can see my complete implementation. There may be possible potential risks here. For example, in overloaded index operator (const function) it returns by value, so read lock assures it's safety. But I can't say the same things for non-const function which returns reference. If you write something like this: T& _refT = mCircularBuffer[0]; It returns as reference, and some other time you are free to change the value of _refT (after leaving the index operator scope). So write lock is unlocked, and as a consequence it's not thread safe so. And also iterators are same as the index operator, not thread safe. I'm looking for much safer implementation. Thanks, regards... template <class T> class CircularBuffer { typedef boost::shared_mutex ReadWriteMutex; typedef boost::shared_lock<boost::shared_mutex> ReadLock; typedef boost::unique_lock<boost::shared_mutex> WriteLock; public: // A const (random access) iterator used to iterate through the circular_buffer typedef typename boost::circular_buffer<T>::const_iterator const_iterator; // A (random access) iterator used to iterate through the circular_buffer typedef typename boost::circular_buffer<T>::iterator iterator; explicit CircularBuffer(unsigned capacity = 10) : m_capacity(capacity) , m_buffer(capacity) { } ~CircularBuffer() { } CircularBuffer(const CircularBuffer& other) : m_capacity(other.m_capacity) , m_buffer(other.m_buffer) { } CircularBuffer& operator= (const CircularBuffer& other) { CircularBuffer temp(other); WriteLock w_lock(rw_mutex); using std::swap; swap(m_capacity, temp.m_capacity); swap(m_buffer, temp.m_buffer); return *this; } // Get the iterator pointing to the beginning of the <code>circular_buffer</code>. iterator begin() { WriteLock w_lock(rw_mutex); return m_buffer.begin(); } // Get the iterator pointing to the end of the <code>circular_buffer</code>. iterator end() { WriteLock w_lock(rw_mutex); return m_buffer.end(); } // Get the const iterator pointing to the beginning of the <code>circular_buffer</code> const_iterator begin() const { ReadLock r_lock(rw_mutex); return m_buffer.begin(); } // Get the const iterator pointing to the end of the <code>circular_buffer</code>. const_iterator end() const { ReadLock r_lock(rw_mutex); return m_buffer.end(); } T& operator [] (unsigned index) { WriteLock w_lock(rw_mutex); return m_buffer[index]; } T operator [] (unsigned index) const { ReadLock r_lock(rw_mutex); return m_buffer[index]; } T& at(unsigned index) { WriteLock w_lock(rw_mutex); return m_buffer.at(index); } T at(unsigned index) const { ReadLock r_lock(rw_mutex); return m_buffer.at(index); } T& front() { WriteLock w_lock(rw_mutex); return m_buffer.front(); } T& back() { WriteLock w_lock(rw_mutex); return m_buffer.back(); } T front() const { ReadLock r_lock(rw_mutex); return m_buffer.front(); } T back() const { ReadLock r_lock(rw_mutex); return m_buffer.back(); } size_t size() const { ReadLock r_lock(rw_mutex); return m_buffer.size(); } bool empty() const { ReadLock r_lock(rw_mutex); return m_buffer.empty(); } bool full() const { ReadLock r_lock(rw_mutex); return m_buffer.full(); } size_t capacity() const { ReadLock r_lock(rw_mutex); return m_buffer.capacity(); } void push_back(const T& item = T()) { WriteLock w_lock(rw_mutex); m_buffer.push_back(item); } void clear() { WriteLock w_lock(rw_mutex); m_buffer.clear(); } private: unsigned m_capacity; boost::circular_buffer<T> m_buffer; // rw_mutex for internal buffer mutable ReadWriteMutex rw_mutex; };

1 0

[Boost.Test] BOOST_CHECK_EQUAL and std::type_info
by Paul S. Strauss 11 Jun '08

11 Jun '08

I can't seem to get BOOST_CHECK_EQUAL with std::type_info instances to compile with g++, version (GCC) 4.1.2 20070925 (Red Hat 4.1.2-33) Here's a simple test case: ------------------ #include <iostream> #include <typeinfo> #define BOOST_TEST_MAIN #include <boost/test/included/unit_test.hpp> std::ostream & operator<<(std::ostream &out, const std::type_info &info) { return out << info.name(); } BOOST_AUTO_TEST_CASE(Foo) { std::cout << "Typeid of int is " << typeid(int) << "\n"; BOOST_CHECK_EQUAL(typeid(int), typeid(float)); } ------------------ g++ complains that there isn't an operator<< for type_info in the BOOST_CHECK_EQUAL, even though it is used just fine on the previous line. Output: g++ -o ti ti.cpp /usr/include/boost/test/test_tools.hpp: In member function ‘void boost::test_tools::tt_detail::print_log_value<T>::operator()(std::ostream&, const T&) [with T = std::type_info]’: /usr/include/boost/test/test_tools.hpp:404: instantiated from ‘std::ostream& boost::test_tools::tt_detail::operator<<(std::ostream&, const boost::test_tools::tt_detail::print_helper_t<T>&) [with T = std::type_info]’ /usr/include/boost/test/utils/wrap_stringstream.hpp:66: instantiated from ‘boost::basic_wrap_stringstream<CharT>& boost::operator<<(boost::basic_wrap_stringstream<CharT>&, const T&) [with CharT = char, T = boost::test_tools::tt_detail::print_helper_t<std::type_info>]’ /usr/include/boost/test/test_tools.hpp:460: instantiated from ‘void boost::test_tools::tt_detail::check_frwd(Pred, boost::wrap_stringstream&, boost::test_tools::const_string, size_t, boost::test_tools::tt_detail::tool_level, boost::test_tools::tt_detail::check_type, const Arg0&, const char*, const Arg1&, const char*) [with Pred = boost::test_tools::tt_detail::equal_impl_frwd, Arg0 = std::type_info, Arg1 = std::type_info]’ ti.cpp:14: instantiated from here /usr/include/boost/test/test_tools.hpp:298: error: no match for ‘operator<<’ in ‘ostr << t’ <List of candidates removed, but type_info isn't one of them> Anyone know why this is? -- Paul S. Strauss pss(a)acm.org

3 3

[Test] custom Exception
by Jarl Friis 11 Jun '08

11 Jun '08

Is there any nice way to achive more detailed information about uncaught custome exceptions in a Boost unit test. I understand that it is possible to register custom exceptions, but that is quite some work when using several 3rd party libraries. I would be even hapier if it was possible to get some stacktrace, or linenumber (within the Boost test file) on which the exception occured. Jarl

2 1

Finding several Files
by German Escallon 11 Jun '08

11 Jun '08

Hello All, I am trying to figure out the path to several files w/ the same name inside a certain directory structure. I was thinking of using the find_file() function from the filesystem library (see below). Unfortunately, the function stops when the first match is encountered. Is there a better way and/or alternative?? Thanks. ~G bool find_file( const path & dir_path, // in this directory, const std::string & file_name, // search for this name, path & path_found ) // placing path here if found { if ( !exists( dir_path ) ) return false; directory_iterator end_itr; // default construction yields past-the-end for ( directory_iterator itr( dir_path ); itr != end_itr; ++itr ) { if ( is_directory(itr->status()) ) { if ( find_file( itr->path(), file_name, path_found ) ) return true; } else if ( itr->leaf() == file_name ) // see below { path_found = itr->path(); return true; } } return false; } ______________________________________________________________________ This email has been scanned by the MessageLabs Email Security System. For more information please visit http://www.messagelabs.com/email ______________________________________________________________________

1 0