- Boost-users - lists.boost.org

Re: [Boost-users] AVL tree insert_equal generating assertion `node_algorithms::unique(to_insert)'
by bsing 16 Oct '09

16 Oct '09

My previous mail got truncated, the completed listing is below: Hi, I have the following piece of code that is generating the above assertion. Could someone tell me what is going wrong: template <typename T> struct my_node_t : public avl_set_base_hook<> { my_node_t(T data): m_data(data) {} template <typename U> friend bool operator< (const my_node_t& lhs, const my_node_t& rhs) { return lhs.m_data < rhs.m_data; }; T m_data; }; typedef my_node_t<int> avlnode_t; typedef avltree<avlnode_t> avltree_t; struct args_t { unsigned int seed[MAX_THREADS]; vector<avlnode_t>* items; avltree_t* tree_ptr; thread_args_t(size_t num_threads, avltree_t* map, vector<avlnode_t>* init_vec, size_t num_entries, unsigned int s) { num_items = num_entries; tree_ptr = map; initial_items = init_vec; for (size_t i = 0; i < num_threads; ++i) { seed[i] = i + s; } for (int i = 0; i < 10; ++i) { initial_items->push_back(i * 100); } for (vector<avlnode_t>::iterator it = initial_items->begin(); it != initial_items->end(); ++it) { tree_ptr->insert_equal(*it); } } ~thread_args_t() { tree_ptr->clear(); delete tree_ptr; delete initial_items; } }; void worker(void* args_ptr) { args_t* args = static_cast<args_t*>(args_ptr); avltree_t* tree = args->tree_ptr; vector<avlnode_t>* values = args->initial_items; int index = 5; //generate_random<int>(my_seed, 0, num_items + 1))) avlnode_t& find_element = *(values->begin() + index); tree->insert_equal(find_element); ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ <-raises exception } int main() { int num_threads = 1, num_entries = 10; vector<avlnode_t>* init_vec = new vector<avlnode_t>; avltree_t* tree = new avltree_t; args_t* args = new args_t(num_threads, tree, init_vec, num_entries); worker(args); delete args; }

2 4

AVL tree insert_equal generating assertion `node_algorithms::unique(to_insert)'
by bsing 16 Oct '09

16 Oct '09

Hi, I have the following piece of code that is generating the above assertion. Could someone tell me what is going wrong: template <typename T> struct my_node_t : public avl_set_base_hook<> { my_node_t(T data): m_data(data) {} template <typename U> friend bool operator< (const my_node_t& lhs, const my_node_t& rhs) { return lhs.m_data < rhs.m_data; }; T m_data; }; typedef my_node_t<int> avlnode_t; typedef avltree<avlnode_t> avltree_t; struct args_t { unsigned int seed[MAX_THREADS]; vector<avlnode_t>* items; avltree_t* tree_ptr; thread_args_t(size_t num_threads, avltree_t* map, vector<avlnode_t>* init_vec, size_t num_entries, unsigned int s) { num_items = num_entries; tree_ptr = map; initial_items = init_vec; for (size_t i = 0; i < num_threads; ++i) { seed[i] = i + s; } for (int i = 0; i < 10; ++i) { initial_items->push_back(i * 100); } for (vector<avlnode_t>::iterator it = initial_items->begin(); it != initial_items->end(); ++it) { tree_ptr->insert_equal(*it); } } ~thread_args_t() { tree_ptr->clear(); delete tree_ptr; delete initial_items; } }; void worker(void* args_ptr) { args_t* args = static_cast<args_t*>(args_ptr); avltree_t* tree = args->tree_ptr; vector<avlnode_t>* values = args->initial_items; int index = 5; //generate_random<int>(my_seed, 0, num_items + 1))) avlnode_t& find_element = *(values->begin() + index); tree->insert_equal(find_element); ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ <-raises exception } int main() { int num_threads = 1, num_entries = 10; vector<avlnode_t>* init_vec = new vector<avlnode_t>; avltree_t* tree = new avltree_t; args_t* args = new args_t(num_threads, tree, init_vec, num_entries);

1 0

Anyone see anything like this??
by Noah Roberts 15 Oct '09

15 Oct '09

In mem_fn.hpp (not sure what is doing this but my bet is bind or function) there is a function dm<R,T>::call that looks like so: template<class U> R const & call(U & u, void const *) const { return (get_pointer(u)->*f_); } The stupid compiler (MSVC++ 8.0) complained that u is an unreferenced formal parameter and then crashed. So I told it to stop giving me that warning in the file that is causing the issue and the problem went away...until I set about trying use the function that I just wrote that caused it. Now I get this: 1>d:\boostvs39\include\boost\bind\mem_fn.hpp(334) : error C4716: 'boost::_mfi::dm<void __thiscall(unsigned int),esi::units::unit_settings>::call<esi::pipeflo::document::unit_pool * const>' : must return a value I can't recreate this in a simple program. Here's the code of the function that I just wrote when this BS started happening: void on_change(wxGridEvent & event) { using namespace esi::units::systems::SI; using namespace esi::pipeflo::document; if (event.GetCol() == 0) return; std::pair< boost::function< std::vector< std::string >() > , boost::function< void(size_t) > > functions[] = { std::make_pair( boost::bind(&unit_settings::get_labels<pressure>, &units) , boost::bind(&unit_settings::set_selection <pressure>, &units) ) , std::make_pair( boost::bind(&unit_settings::get_labels<length>, &units) , boost::bind(&unit_settings::set_selection<length>, &units) ) , std::make_pair( boost::bind(&unit_settings::get_labels<mass_flow>, &units) , boost::bind(&unit_settings::set_selection <mass_flow>, &units) ) , std::make_pair( boost::bind(&unit_settings::get_labels <dynamic_viscosity>, &units) , boost::bind(&unit_settings::set_selection <dynamic_viscosity>, &units) ) , std::make_pair( boost::bind(&unit_settings::get_labels <mass_density>, &units) , boost::bind(&unit_settings::set_selection <mass_density>, &units) ) }; std::vector< std::string > labels = functions[event.GetRow()].first (); std::vector< std::string >::iterator fit = std::find(labels.begin(), labels.end(), grid->GetCellValue(event.GetRow(), event.GetCol())); assert(fit != labels.end() && "Somehow the user selected a unit that doesn't exist."); if (fit != labels.end()) { functions[event.GetRow()].second(std::distance(labels.begin(), fit)); } } The namespace esi::units::systems::SI has a "using boost::units::si" in it. I am not really expecting anyone can help me but I thought I'd try anyway. Surely other people have been stabbed in the face by this terribly buggy compiler.

3 3

Boost::Asio -- Is this safe?
by Dan Eaton 15 Oct '09

15 Oct '09

I have a situation where I have two (or more) nodes that need to periodically exchange data in a symmetric fashion - i.e. one needs to send the other a buffer, and vice versa. Currently I implement this as two async operations: for(neighboring nodes) { async_write(...) async_read(...) } m_ioService.run() There exists a tcp socket connection between each of the neighbors. The handlers associated with both read and write merely do error checking. Is this okay, or is there a issue with this approach? In general, it seems to be working fine, but someone brought a bug to my attention whereby they were getting "Transport endpoint not connected." errors happening (under a heavily loaded network, if that makes any difference, but I'm not using any timeouts). Thanks! Dan Eaton

2 1

Re: [Boost-users] Boost::Asio -- Is this safe?
by Dan Eaton 15 Oct '09

15 Oct '09

PS. I should have mentioned that each worker maintains a socket *to* and *from* its partner. I have a situation where I have two (or more) nodes that need to periodically exchange data in a symmetric fashion - i.e. one needs to send the other a buffer, and vice versa. Currently I implement this as two async operations: for(neighboring nodes) { async_write(...) async_read(...) } m_ioService.run() There exists a tcp socket connection between each of the neighbors. The handlers associated with both read and write merely do error checking. Is this okay, or is there a issue with this approach? In general, it seems to be working fine, but someone brought a bug to my attention whereby they were getting "Transport endpoint not connected." errors happening (under a heavily loaded network, if that makes any difference, but I'm not using any timeouts). Thanks! Dan Eaton _______________________________________________ Boost-users mailing list Boost-us...<http://groups.google.com/groups/unlock?_done=/group/boost-list/browse_threa…>@lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

1 0

Directory monitor for Windows and Linux
by Boris Schaeling 15 Oct '09

15 Oct '09

I've updated the directory monitor which I had developed about a year ago: http://www.highscore.de/boost/dir_monitor.zip It now supports Windows and Linux. The Linux version is based on inotify (available since kernel version 2.6.13). The directory monitor supports blocking and asynchronous I/O operations. You use it like any other Boost.Asio I/O object. There are two test cases to verify that the directory monitor works and to provide a minimum documentation. As there are only four methods provided (add_directory, remove_directory, monitor and async_monitor) it should be pretty easy to use this Boost.Asio extension. If anyone wants to add support for more platforms (like Solaris) just drop me a line! Boris

1 0

Re: [Boost-users] Boost::ASIO, Signal handlers
by Dan Eaton 15 Oct '09

15 Oct '09

Nothing, other than to log that the signal was received! There appears to be an intermittent bug in my application, where I get the error " Transport endpoint not connected." which hints at sockets closing unexpectedly (I'm definitely not closing them accidentally!) On Thu, Oct 15, 2009 at 11:43 AM, Dan Eaton <daniel.ea...<http://groups.google.com/groups/unlock?_done=/group/boost-list/browse_threa…>@acceleware.com> wrote: > I was thinking to simply call std::signal() on my own, after the > construction of io_service. Will that work, and is there a better way? I'm assuming that you're going to implement a handler for SIGPIPE, and out of curiosity, what will your handler will do? Jon _______________________________________________ Boost-users mailing list Boost-us...<http://groups.google.com/groups/unlock?_done=/group/boost-list/browse_threa…>@lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

2 1

Boost::ASIO, Signal handlers
by Dan Eaton 15 Oct '09

15 Oct '09

I'm using the ASIO component of boost 1.39.0, and noticed that it uses std::signal() to ignore SIGPIPE by default. Can you confirm that this is indeed the case? How do I prevent this behavior? (and therefore allow for my own signal handler) I was thinking to simply call std::signal() on my own, after the construction of io_service. Will that work, and is there a better way? Thank you, Dan Eaton

2 1

Boost.Asio -- combining sync read with async read
by Sameer Parekh 15 Oct '09

15 Oct '09

Thanks Igor and Jon, everything seems to be working now! yours, -sameer

1 0

Question on intrusive hooks
by bsing 15 Oct '09

15 Oct '09

I'm trying to come to grips with using the boost intrusive containers, but I don't think I understand how the lifetime of hooks need to be managed. This piece of code fails the !hook.is_linked() assertion: Any pointers? Also, is there is a non-intrusive AVL tree container as part of boost? struct my_node_t { int i_; my_node_t(int i): i_(i) {} friend bool operator<(const my_node_t& lhs, const my_node_t& rhs) { return lhs.i_ < rhs.i_; } friend ostream& operator<< (ostream& os, const my_node_t& n) { return os << n.i_; } }; int main() { avltree<my_node_t> tree; vector<my_node_t> vec; for (int i = 0; i < 10; ++i) vec.push_back(my_node_t(i)); for (int i = 0; i < 10; ++i) tree.insert_unique(vec[i])); copy(tree.begin(), tree.end(), ostream_operator<my_node_t>(cout, " "); }

2 1