- Boost-users - lists.boost.org

Re: [Boost-users] [Graph] Adjacency list scalability with a Dijks tra algorithm
by Florian.PONROY＠fr.thalesgroup.com 26 Mar '09

26 Mar '09

Hi Juan, Thanks for your answer. I definitly agree with the map overhead, I will try to better understand how the property map works in order to implement your solution. Florian -----Message d'origine----- De : boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org]De la part de Juan Antonio Farré Basurte Envoyé : mercredi 25 mars 2009 19:27 À : boost-users(a)lists.boost.org Objet : Re: [Boost-users] [Graph] Adjacency list scalability with a Dijks tra algorithm Hi, It's as easy as instantiating the adjacency_list template with mapS for the vertex list selector. This is the second template argument. For example, typedef boost::adjacency_list<boost::vecS,boost::mapS> myGraph; Anyway, remember that a map is more expensive than a vector in terms of both space and time (access time to an element is logarithmic). Another option is that you stay with the default vector for the vertex list, and you use a property to store the vertex id. This way, your vertex 4242 would have descriptor 6 and you'd store 4242 as its id-property value. Cheers, Juan > Hi Andrew, > > Would you please tell me which part of the documentation I should rely on to use a std::map as you'd suggested? I should actually admit I'm not yet > familiar with the BGL :-) > > > -----Message d'origine----- > De : boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org]De la part de Andrew Sutton > Envoyé : mercredi 25 mars 2009 17:25 > À : boost-users(a)lists.boost.org > Objet : Re: [Boost-users] [Graph] Adjacency list scalability with a Dijkstra algorithm > > > > > Here is the problem: I would have imagined that, in my case, > boost::num_vertices(graph) would have returned 6, but it actually returns > the higher vertex number + 1, that is... 4243 ! So each of my vector has a > size of 4243 * sizeof(tVertex) bytes... Is there a way to handle that? I > probably misuse the API, but it would be better if the container would be > dimensioned to the actual number of vertices, since I am in a CPU and memory > constraint environment... > > > > You're confusing the notions of "index" and "descriptor". With the vertex set using vecS, the descriptor of a vertex is the same as its index -- > hence the confusion. Therefore, when you call add_edge(4242, 1, ...), the graph thinks you mean the vertex whose descriptor is 4242 and resizes > itself accordingly. > > You have to remember - whatever the examples may indicate - that you should never actually use a literal value and descriptors interchangeably. This > isn't documented very well, either. > > You could use a std::map to associate your own identifiers with vertex descriptors. > > > Andrew Sutton > andrew.n.sutton(a)gmail.com <mailto:andrew.n.sutton@gmail.com> > > > _______________________________________________ > Boost-users mailing list > Boost-users(a)lists.boost.org > http://lists.boost.org/mailman/listinfo.cgi/boost-users

1 0

unicode, C++0x, UTF-16
by Brent Arias 26 Mar '09

26 Mar '09

I see that wstring_convert and codecvt_utf8_utf16 are in the current C++0x spec. Are there plans to backfill boost with implementations for these? I'm also curious why the critique of wstring_convert (http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2005/n1750.html) was not incorporated for C++0x. Thoughts? I've also spotted a comment in boost codecvt_utf8_facet saying: // note the following code will generate on some platforms where // wchar_t is defined as UCS2. The warnings are superfluous as // the specialization is never instantitiated with such compilers. Does that mean that boost.codecvt_utf8_facet is capable of acting as a UTF-8 / UCS2 converter on a Windows box, where wchar_t is 2 bytes in size? Brent Arias Software Architect Lumension Security(tm) Tel ~ (602) 692-8088 Fax ~ (480) 970-6323 brent.arias(a)lumension.com<mailto:brent.arias@lumension.com>

2 1

[fusion] complexity of filter_view
by er 26 Mar '09

26 Mar '09

Hi All, If say vector_type has N elements and predicate is true only for one of them. O(N)? typedef filter_view<vector_type const, predicate > view_t O(1)? view_t view(v);

2 1

Steven Watanabe template profiler
by Joel Falcou 26 Mar '09

26 Mar '09

A while back, I claimed having problems using Steven template profiler. I'm ashamed to say that the problem was that gcc is actually localized. So the warning I got in the filtered output was : "./test.hpp:17: attention : ‘int template_profiler::f(int)’ is deprecated (declared at ./template_profiler.hpp:50) " So I guess I have to edit the regex in postprocess ? I am not a regex guru so If Steven walk by, could you give me some directions to fix this on my end. -- ___________________________________________ Joel Falcou - Assistant Professor PARALL Team - LRI - Universite Paris Sud XI Tel : (+33)1 69 15 66 35

2 1

compile-time check that a member function exists
by John Pretz 25 Mar '09

25 Mar '09

Hi, I'm pretty new to boost. I want to write some code with a compile-time check whether a member function exists or not. The code below shows kinda what I need except I'd obviously like to avoid a specialized template for each class that I use. Can anybody point me in the right direction? Thanks, John **************************************** #include <iostream> using namespace std; struct WithMemberGo { void Go() const {cout<<__PRETTY_FUNCTION__<<" going...."<<endl;} }; struct WithoutGo { }; template <class T> void CallGoIfItExists(const T& t){ t.Go(); } // I want to avoid having to specialize this function for every class without the // right member function. template <> void CallGoIfItExists<WithoutGo>(const WithoutGo&){ cout<<"not going anywhere...."<<endl; } int main(){ WithMemberGo with; CallGoIfItExists(with); WithoutGo without; CallGoIfItExists(without); } -- John Pretz -- P-23, MS H803 Los Alamos National Lab Los Alamos, NM 87545 Phone: (505)665-5847 Fax: (505)665-4121

5 10

Re: [Boost-users] [Graph] Adjacency list scalability with a Dijks tra algorithm
by Florian.PONROY＠fr.thalesgroup.com 25 Mar '09

25 Mar '09

Hi Andrew, Would you please tell me which part of the documentation I should rely on to use a std::map as you'd suggested? I should actually admit I'm not yet familiar with the BGL :-) -----Message d'origine----- De : boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org]De la part de Andrew Sutton Envoyé : mercredi 25 mars 2009 17:25 À : boost-users(a)lists.boost.org Objet : Re: [Boost-users] [Graph] Adjacency list scalability with a Dijkstra algorithm Here is the problem: I would have imagined that, in my case, boost::num_vertices(graph) would have returned 6, but it actually returns the higher vertex number + 1, that is... 4243 ! So each of my vector has a size of 4243 * sizeof(tVertex) bytes... Is there a way to handle that? I probably misuse the API, but it would be better if the container would be dimensioned to the actual number of vertices, since I am in a CPU and memory constraint environment... You're confusing the notions of "index" and "descriptor". With the vertex set using vecS, the descriptor of a vertex is the same as its index -- hence the confusion. Therefore, when you call add_edge(4242, 1, ...), the graph thinks you mean the vertex whose descriptor is 4242 and resizes itself accordingly. You have to remember - whatever the examples may indicate - that you should never actually use a literal value and descriptors interchangeably. This isn't documented very well, either. You could use a std::map to associate your own identifiers with vertex descriptors. Andrew Sutton andrew.n.sutton(a)gmail.com <mailto:andrew.n.sutton@gmail.com>

2 1

Statechart problem
by Iwan Sanders 25 Mar '09

25 Mar '09

LS, I've implemented a state machine (synchronous) with a couple of states and their events. Its purpose is to implement an USDT CAN stack. Because the lower level driver (made that too) passes the incoming CAN messages through a callback function I needed to derive one state in the state machine from: a) the expected boost state base b) my own interface which handles the callback for new messages (observer) The state machine is performing well with one exception, hence this email. The callback function is a member of a struct which is derived from a & b and it is called by the CAN driver. But then when I want to notify the state that the machine is currently in with a post_event this event seems to be discarded even though it should react on it (the machine state has transitions defined for this event and the current state is that state). I also tried using the transit call but that doesn't do anything either. It is just this one member-function that is the problem and before rewriting the state-machine to an asynchronous state machine I wanted to consult some other users. Regards, Iwan Sanders

3 7

Re: [Boost-users] [asio] Disable nagle algorithm on Windows XP
by 张亚霏 25 Mar '09

25 Mar '09

try native WSA API "setsockopt" with the TCP_NODELAY option. if it still does not work, there may be something wrong with your windows xp. or debug the asio ------------------ best wishes to you 张亚霏 ------------------ Original ------------------ From: "Tobias J?hnel"<tjaehnel(a)gmail.com>; Date: Wed, Mar 25, 2009 10:58 PM To: "boost-users"<boost-users(a)lists.boost.org>; Subject: Re: [Boost-users] [asio] Disable nagle algorithm on Windows XP Hi Igor, thanks for your reply. I guess you mean setting non_blocking_io to false. Tried to enter those lines, but still have no change: boost::asio::socket_base::non_blocking_io command(false); socket.io_control(command); Tobias On Wed, Mar 25, 2009 at 3:07 PM, Igor R <boost.lists(a)gmail.com> wrote: >> I'm already using blocking sockets as suggest in above document. > > According to that document, you should not enocounter this problem > when using blocking sockets. Probably your socket is not blocking? > To make a socket blocking you should write something like this: > boost::asio::socket_base::non_blocking_io command(true); > socket.io_control(command); > _______________________________________________ > Boost-users mailing list > Boost-users(a)lists.boost.org > http://lists.boost.org/mailman/listinfo.cgi/boost-users > _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

2 1

Re: [Boost-users] [asio] ip::tcp::socket - differences between send &write_some
by 张亚霏 25 Mar '09

25 Mar '09

after referring the source code, i hardly found difference between ip::tcp::socket::send & ip::tcp::socket::write_some, but about some arguments. "socket_ops::send" directly invokes system call:"sendmsg" to send message. ------------------ best wishes to you 张亚霏 ------------------ Original ------------------ From: "Lennyk"<lennyk430(a)gmail.com>; Date: Wed, Mar 25, 2009 04:59 PM To: "boost-users"<boost-users(a)lists.boost.org>; Subject: [Boost-users] [asio] ip::tcp::socket - differences between send &write_some Hi, What is the difference in implementation between the ip::tcp::socket::send & ip::tcp::socket::write_some on Linux sys sockets? What kind of system calls are used in both calls? Thanks! _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

1 0

[asio] Disable nagle algorithm on Windows XP
by Tobias Jähnel 25 Mar '09

25 Mar '09

Hi, I'm developing a TCP-Client using boost::asio and discovered that ACKs to responses from the server are sent with some delay (~200ms) and I get the data not before this ACK has been sent. When I disable nagle in registry as described by Microsoft it works fine. (See: http://support.microsoft.com/kb/823764) To avoid those registry hacks I want to find another way to get rid of this latency. I'm already using blocking sockets as suggest in above document. I also tried to set the no_delay option but this has no effect at all. boost::asio::ip::tcp::no_delay option(true); socket.set_option(option); == Code Snippets == I use the boost asio library as follows: // open connection (from synchronous TCP daytime client example) boost::asio::io_service io_service; tcp::resolver resolver(io_service); tcp::resolver::query query(ipaddress, port); tcp::resolver::iterator endpoint_iterator = resolver.resolve(query); tcp::resolver::iterator end; boost::system::error_code error = boost::asio::error::host_not_found; while (error && endpoint_iterator != end) { socket.close(); socket.connect(*endpoint_iterator++, error); } // send some data boost::asio::write(socket, boost::asio::buffer(package, size)); // reception is done inside a thread while(true) { // receive data size_t length = socket.read_some(boost::asio::buffer(resp), error); if (error == boost::asio::error::eof) { // Connection closed cleanly by peer. // ... do sth. } else if (error) { // Some error has happened // ... do sth. } // ... process received data and notify main thread to send the next piece of data } ======= Thank you for all suggestions. Tobias

2 2