- Boost-users - lists.boost.org

Re: [Boost-users] [Graph] betweenness_centrality - output completed percentage?
by The Maschine 24 Jan '13

24 Jan '13

Hi Nick, Thanks a lot for both comments! We currently use an "approximation/sub sampling" algorithm done 10years ago by a colleague but its take 10 times more than BGLs and Im not sure its statistically verified. The paper seems really good. About the brandes_betweenness_centrality_impl(), I suppose you are implying that I have to change the boost.graph header file directly? Right? Best, Tasos > > Message: 5 > Date: Wed, 23 Jan 2013 20:54:16 -0500 > From: Nick Edmonds <ngedmond(a)cs.indiana.edu> > To: boost-users(a)lists.boost.org > Subject: Re: [Boost-users] [Graph] betweenness_centrality - output > completed percentage? > Message-ID: <F264944D-B0A4-4FB6-BDF0-635E9DCCC199(a)cs.indiana.edu> > Content-Type: text/plain; charset="us-ascii" > > Hi Tasos, > > The outer loop of betweenness centrality is a loop over all the vertices > in the graph with each used as the source of a shortest paths computation. > If you want to see how far along the algorithm is you can just instrument > the "for (boost::tie(s, s_end) = vertices(g); s != s_end; ++s)" in > brandes_betweenness_centrality_impl(). Note that you don't actually need > to count shortest paths from *every* vertex in the graph to get a good > approximation of betweenness centrality, check out this paper [1]. > Sampling the vertices in the graph and counting shortest paths from a > subset of them should significantly speed up betweenness centrality if a > (relatively good) approximation will work for your application. > > Cheers, > Nick > > [1] http://www.cc.gatech.edu/~bader/papers/ApproxBC.html > > On Jan 22, 2013, at 8:04 AM, Tasos wrote: > > > Hello all, > > > > I calculate the betweenness-centrality of a large graph > > (V:150k, E:500k, boost::adjacency_list<boost::vecS, boost::vecS, > > boost::undirectedS,) with boost::brandes_betweenness_centrality > > and I can find how to get a indication of the 'completed' > > part of the algorithm. > > > > For example this graph took 1h:38m (single thread 10.8 iMac 3.4ghz) > > to complete but in order to be able to know the execution time > > of larger graph im a bit lost... > > > > On "one to all" BFS you know how many vertices you > > have left but this algorithm seems to me lacking a custom > > visitor or something in order to output some data. > > > > Any ideas as Im really new to boost.graph? > > > > Thanks a lot, > > Tasos > > > > _______________________________________________ > > Boost-users mailing list > > Boost-users(a)lists.boost.org > > http://lists.boost.org/mailman/listinfo.cgi/boost-users > >

1 0

Re: [Boost-users] Range library suggestion
by John M. Dlugosz 24 Jan '13

24 Jan '13

On 1/24/2013 3:59 AM, boost-users-request-at-lists.boost.org |Boost/Allow to home| wrote: > >> >I just added my second metafunction of the form is_range_of_X<T> for use >> >with enable_if, and although the first one took some fiddleing to pass >> >both compilers I was using, the second was a clone of the first with the >> >name and type changed. So it really ought to be a general purpose >> >is_range_of<Whatever, T> metafunction. > I assume it it's simply something like > is_same<typename mpl::eval_if< is_range<T>, range_value<T>, void>::type, > Whatever> I ended up with: template <typename Flag, typename T> struct is_range_of_MYTHING2 : boost::mpl::false_ { }; template <typename T> struct is_range_of_MYTHING2<boost::mpl::true_, T> { typedef typename boost::range_value<T>::type val_type; // chokes here if T is not a range typedef typename is_convertible<val_type, MYTHING>::type type; }; template <typename T> struct is_range_of_MYTHING : is_range_of_MYTHING2<typename boost::has_range_iterator<T>::type, T> { }; I'm familiar enough with type_traits, but not with all of Boost MPL. Is eval_if a lazy-evaluation for short circuiting the AND of the two tests? ---John

1 0

boost asio disign
by Christer Borgqvist 24 Jan '13

24 Jan '13

Hello, how can I design an application that will have a http server like HTTP Server 3 example, but also x number of tcp clients connecting to a server, client must connect to the server before the http server is active, if client is idle with a timer events send an 'I'm here' in every 10 seconds, i tink i have to put the tcp client on a list or ?. Data is received by the http server get the client from list, if non available wait for x amount of time, if time elapsed return incomming data, else, parsing on incomming data, send to tcp server, parse response data put the client back on the list, send response on http, when sending data via tcp client reset the timer. I have the http server working but how do I do with the tcp client, design the threads, should there be only one io_service?, I dont now, im lost! Appreciate all the help. /Christer

3 3

[Graph] betweenness_centrality - output completed percentage?
by Tasos 24 Jan '13

24 Jan '13

Hello all, I calculate the betweenness-centrality of a large graph (V:150k, E:500k, boost::adjacency_list<boost::vecS, boost::vecS, boost::undirectedS,) with boost::brandes_betweenness_centrality and I can find how to get a indication of the 'completed' part of the algorithm. For example this graph took 1h:38m (single thread 10.8 iMac 3.4ghz) to complete but in order to be able to know the execution time of larger graph im a bit lost... On "one to all" BFS you know how many vertices you have left but this algorithm seems to me lacking a custom visitor or something in order to output some data. Any ideas as Im really new to boost.graph? Thanks a lot, Tasos

2 1

Boost.Process and STARTUPINFOEX
by Nelson Teixeira 23 Jan '13

23 Jan '13

Hello, First, if this is not the correct place to ask, please point me an appropriate forum/maillist about Boost.Process as I couldn't find any in their site. Thanks. I'm trying to use Boost.Process library in windows XP. I installed Boost libraries version 1.52, and used Boost.Process from this page: http://www.highscore.de/boost/process0.5/index.html however, when I made the includes in my test program, I get this error: C:\boost/boost/process/windows/executor.hpp:120:5: error: 'STARTUPINFOEX' does not name a type C:\boost/boost/process/windows/executor.hpp: In constructor 'boost::process::windows::executor::executor()': C:\boost/boost/process/windows/executor.hpp:24:24: error: 'EXTENDED_STARTUPINFO_PRESENT' was not declared in this scope C:\boost/boost/process/windows/executor.hpp:30:23: error: 'startup_info_ex' was not declared in this scope C:\boost/boost/process/windows/executor.hpp:34:9: error: 'STARTUPINFOEX' was not declared in this scope I can't find this STARTUPINFOEX. At most I have STARTUPINFO structure. Do I have to install anything else ? Can anyone help please ?/ Thanks -Nelson ------------------------------------------------- Blog: Além do Óbvio http://www.alemdoobvio.com

4 3

intrusive_ptr vs shared-ptr
by ilya shlyakhter 23 Jan '13

23 Jan '13

Dear Boosters, The documentation of intrusive_ptr says "As a general rule, if it isn't obvious whether intrusive_ptr better fits your needs than shared_ptr, try a shared_ptr-based design first." but does not explain why. If the classes involved are all my own and deriving them all from a base class that implements reference-counting is not a problem, what are some reasons to prefer shared_ptr to intrusive_ptr? Thanks, Ilya

5 8

Range library suggestion
by John M. Dlugosz 23 Jan '13

23 Jan '13

I just added my second metafunction of the form is_range_of_X<T> for use with enable_if, and although the first one took some fiddleing to pass both compilers I was using, the second was a clone of the first with the name and type changed. So it really ought to be a general purpose is_range_of<Whatever, T> metafunction. I would suggest that such a thing be added to the Range library. It is meant for overloaded forms that operate on ranges, and you don't want to accept just any ol' thing because, perhaps, you want other overloads or to get better error messages for bad arguments. A typical example would be something like: void insert (const C& x); template <typename R> typename enable_if<typename is_range_of<C,R>::type, void>::type insert (const R& range) { ⋯ } In my case, I check whether R is indeed a Range (something that appears at first to be missing from the library too, but has_range_iterator is probably what is really meant by that) and the underlying element is convertible to the desired type. In a general-purpose library you might need to put more thought into what tests are so common as to be worth reusing at that level. I think having convertable-to-C (where you just give C) and another one which lets you pass in another metafunction for matching the element would allow reuse of experly written common code that takes care of nuances across compilers, e.g. choking on value_type if the outer is NOT a range, so you need to ensure short-circuit evaluation of the meta-expression. —John

2 1

[mpl] Suggestion on handle protocol messages
by Daniele Barzotti 23 Jan '13

23 Jan '13

Dear, I need to handle a protocol defined with Protocol Buffers. My messages are defined as: enum EMessages { E_MSG_METHOD_CONNECT = 0x8001, E_MSG_EVENT_CONNECT = 0xA001, ... } struct MsgHeader { long sessionRef; long transactionId; long status; } struct MSG_METHOD_CONNECT { Messages opCode; MsgHeader header; .. other fields .. } Now, I defined an interface and a template class to add a level of indirection.. class IMessage { virtual INT getOpCode() = 0; virtual STRING getName() = 0; virtual size_t getSize() = 0; .... } template<class MESSAGE> class IMessageImpl : public IMessage { protected: MESSAGE m_Message; ///< The Message Implementation public: virtual MESSAGE& getMessage() = 0; }; Where I'll use IMessageImpl<MSG_METHOD_CONNECT> MsgConnect; Now, the problem in this situation is to build the correct message class on the receiving part, knowing only the Message opCode. I would like to manage the messages with a type list or map, but I'm a newbie in MPL... Is there a way to define something like this: boost::mpl::map<int, MESSAGE> where the key is the Message opCode and the MESSAGE is the message type? The idea was born reading this article: http://blog.florindinu.ro/posts/message-dispatching-in-c-part-1-type-lists.… Thanks in advance, Daniele.

1 0

Boost and OMPL : access graph vertex
by Dan 23 Jan '13

23 Jan '13

Hello, I'm using OMPL (the Open Motion Planning Library) that makes use of Boost. The problem I've got is that I need to access the content of the vertexes of a graph which is returned by an OMPL function. In particular, the OMPL function is const Graph& g = planner->as<ompl::geometric::PRM>()->getRoadmap(); Now, given the graph g, and a vertex v, I need to read the state contained in g[v] and copy it to an OMPL variable. I'd need something like const ompl::base::State* state = FUNCTION_GET_STATE(g,v) ; This way I would be able to extract from variable state the components x,y,z I'm interested in. Is there any way to implement the FUNCTION_GET_STATE ? Thank you! Dan -- View this message in context: http://boost.2283326.n4.nabble.com/Boost-and-OMPL-access-graph-vertex-tp464… Sent from the Boost - Users mailing list archive at Nabble.com.

1 0

Re: [Boost-users] boost::bind/boost::function and fast, delegates
by John M. Dlugosz 23 Jan '13

23 Jan '13

On 1/22/2013 4:34 AM, boost-users-request-at-lists.boost.org |Boost/Allow to home| wrote: > And if you still wish to write such a lib(s) take a look at this article: > > http://www.codeproject.com/Articles/7150/Member-Function-Pointers-and-the-F… Coincedentely, that article references my project that I cited on another thread (shared pointers): > For Microsoft's (and Intel's) ugly, non-standard 12 byte |virtual_inheritance| pointers, > yet another trick is used, based on an idea invented by John Dlugosz. To pass bound member functions to legacy code (including OS API) that expects a genuine function pointer, I generated a thunk at run-time and used a compiler trick to get the information I needed regarding the 'this' adjustment. It is certainly the _fastest_ possible; but I did not elaborate on a full "bind" feature like Boost (and std) eventually came up with. In fact, I simplified my own code so that all the arguments to pass were just members of one structure, so I only worried about the case of binding to a member function with one argument. It didn't automatically capture smart-pointer objects for the binding, but I handled that separately since it was already being done for "windows" owned by the OS's reference to raw pointers; I had my own fake +1 count as part of the object, and Window objects were my primary interest. As I recall, the C++11 lambda spec provides for a primitive C function pointer to be coughed up, but I've not yet used a compiler that can do that.

1 0