May 2013 - Boost-users - lists.preview.boost.org

[asio] schedule work on all threads of a single io_service::run() pool?
by Anthony Foiani 18 May '13

18 May '13

Greetings -- Is there a way to ask the io_service to distribute a particular handler to every thread that is running the service? My use case is per-thread watchdogs, and I need the worker threads to ping the watchdog on some regular interval (e.g., every 5 seconds). I didn't see anything like this in the docs, and I didn't see anything that looked promising when I looked through the source. (I'll freely admit that the ASIO source is beyond my ability to comprehend it without much more study, however.) I can approximate it by having my worker threads do something like this (pseudocode): while ( true ) { boost::system::error_code ec; io_service::poll_one( ec ); if ( ec == stopped ) break; watchdog::ping(); sleep( 500ms ); // trying to balance lag and wakeups } But that has the obvious problems of lag and more wakeups than I really need. To make this hack work, I think I need some calls that probably don't exist; instead of that "sleep" call, I would like to be able to do something like a timed wait on a condition variable: io_service::all_threads_cond.timed_wait( 5s ); That way, I would only wake up every 5 seconds (to feed the watchdog), unless the io_service sent a "notify_one" to the condition variable and this thread happened to be picked. Extra pings aren't much of an issue, but it would be nice to minimize them. Anyway. I'll continue digging, but if anyone has suggestions, I'd love to hear them. Thanks in advance! Best regards, Anthony Foiani

2 5

[smart_ptr] Migrating from boost::shared_ptr<> to std::shared_ptr<>
by Szymon Gatner 18 May '13

18 May '13

Hi, with recently acquired (semi) C++11 compiler (YEY!) I started to think which of BOOST functionality could be replaced by standard counterparts. First obvious candidates are smart pointers. Thing is, there is lots of functions that take boost::shared_ptr<> as arguments (or return them) and it is just not easy to do do. My questions: - does it even make sense to switch to std::shared_ptr<> (should new functions use std instead of boost) - is there a way to easily convert one to another - what are (if any) differences between std and boost versions Regards, Szymon Gatner

2 1

(no subject)
by Yeni Lora 18 May '13

18 May '13

Hi I have a problem with my code I using mpi. The error is with mca_btl_tcp_frag_recv, and I don´t know what mean... The error complete [../../../../../../ompi/mca/btl/tcp/btl_tcp_frag.c:216:mca_btl_tcp_frag_recv] mca_btl_tcp_frag_recv: readv failed: Connection timed out (110) and [btl_tcp_frag.c:216:mca_btl_tcp_frag_recv] mca_btl_tcp_frag_recv: readv failed: Connection reset by peer (104) I need some help...

4 3

[lockfree] overwriting push?
by Rich E 18 May '13

18 May '13

I'm using boost::lockfree::spsc_queue in an audio graph to allow audio input samples written from one thread to be read from another. This currently looks like: #include <boost/lockfree/spsc_queue.hpp> class RingBuffer { public: RingBuffer() : mLockFreeQueue( 0 ) {} RingBuffer( size_t size ) : mLockFreeQueue( size + 1 ), mSize( size ) {} ~RingBuffer() {} //! pushes to buffer, overwriting oldest samples void write( const float *samples, size_t count ) { if( count > mSize ) { count = mSize; } size_t numPushed = mLockFreeQueue.push( samples, count ); if( count > numPushed ) { size_t numLeft = count - numPushed; // ???: is there a more efficient way to overwrite? float old; for( int i = 0; i < numLeft; i++ ) { mLockFreeQueue.pop( old ); } numPushed = mLockFreeQueue.push( &samples[numPushed], numLeft ); assert( numPushed == numLeft ); } } size_t read( float *samples, size_t count ) { if( count > mSize ) { count = mSize; } return mLockFreeQueue.pop( samples, count ); } private: boost::lockfree::spsc_queue<float> mLockFreeQueue; size_t mSize; }; If there is not enough space in the write call, I sequentially pop off elements until there is enough. Is there a better way to do this? It seems like there should be a mechanism for just moving the read head forward (atomically) without reading anything, but then I am not sure I completely understand the implications of that on the lockfree model. I'm also not 100% sure this is adhering to the lockfree model, so please inform me if/how it isn't if you know. Thank you for your time, Rich

2 2

[BGL] assert( source(* (out_edges(v, undigraph).first), undigraph) == v )
by Andreas Fabri 17 May '13

17 May '13

Hello, When I have an undirected graph, for edge_descriptor source and target still are defined. That is stated here: http://www.boost.org/doc/libs/1_53_0/libs/graph/doc/graph_concepts.html My question. Is edge_descriptor ed = * out_edges(v,undigraph).first; Is there a requirement for graph implementations that source(ed,g) == v ? best regards, andreas -- Andreas Fabri, PhD Chief Officer, GeometryFactory Editor, The CGAL Project phone: +33.492.954.912 skype: andreas.fabri

2 1

which library to use:cleaning up singletons
by Littlefield, Tyler 17 May '13

17 May '13

hello all: I have a fairly large code base that has a lot of modules and systems. The problem is that I have a main "engine" object that holds pointers to a lot of created modules and components. I'm trying to find a more dynamic way to do this, and I thought boost might be able to help. As a result, I have a couple of questions. First, I do a lot of manual adding of features. For example, when I add a new command I have to add a engine->commands.AddCommand(new CMDFoo()); in an initialization method. Does boost provide some sort of mechenism for dynamically handling this? I also plan on splitting up the main "engine" object and setting up the external modules as singletons. They act alone, but again I'd like to dynamically initialize these. Is there a way on entry to dynamically call specific singleton initialization methods, then at exit call their release methods? I'd like to be able to set some sort of trait on these objects at compile-time that the runtime can use, or somehow generate the calls at compile-time. Thanks in advance for any info, -- Take care, Ty http://tds-solutions.net He that will not reason is a bigot; he that cannot reason is a fool; he that dares not reason is a slave. Sent from my Toaster (tm).

1 0

[Boost.Atomic] Wait-free multi-producer queue example bug?
by Ioannis Papadopoulos 16 May '13

16 May '13

Hi, In http://www.boost.org/doc/libs/1_53_0/doc/html/atomic/usage_examples.html#bo… I see that pop_all_reverse() is implemented as: node * pop_all_reverse(void) { return head_.exchange(0, boost::memory_order_consume); } Shouldn't the memory order be memory_order_acquire in this example? If I'm not mistaken, at least an acquire is required to see all the writes from the thread that did the release. GCC 4.8 also complains that memory_order_consume is not a correct memory ordering for exchange. Thanks

1 0

[Graph] Problem with large vertex ids
by Stephen Woodbridge 16 May '13

16 May '13

Hi all, I'm trying to understand/fix some bugs in pgRouting that use the boost libraries. I know C not not C++, so I can understand most of the more C-ish parts. We have an old bug that I believe is caused when we add an edge and it has large vertex ids or when there is a large number between the min and max vertex numbers. We currently identify the min id number and down shift the ids by that, but that does not help if I add an edge (1)->(100000000). I could renumber all the nodes and then un-renumber them when done if I have to. Questions: 1. is this a known limitiation in Boost graph? 2. is there a different type of adjacency list mechanism that does not have this issue 3. currently the code crashes the database backend, ideally I would like to catch any issues in the C++ wrapper, free memory and return an error code to the C caller so we don't kill the server or leak lots of memory. Here is a link to our boost wrapper code: https://github.com/pgRouting/pgrouting/blob/sew-devel-2_0/src/dijkstra/src/… We use that pattern for a lot of functions so if there is a better pattern, I might be able to update most(all?) of the functions that use this. I could really use some help resolving this issue. Meanwhile, I'm off to write a C main() to run this outside of the database and in gdb to see whats happening. Thanks, -Steve, trying to release pgRouting 2.0

3 4

Interprocess : vectors of vectors
by Oodini 16 May '13

16 May '13

Hello, I'd like to share a vector of vectors (jagged 2D array). The data come from a deserialization. I know the size of the "external" vector. The sizes of the "internal" vectors are discovered during deserialization. I read : http://www.boost.org/doc/libs/1_52_0/doc/html/interprocess/quick_guide.html… and regarding this line : managed_shared_memory segment(create_only, "MySharedMemory", //segment name 65536); how to set the size (3rd argument), when considering my use case ? Even with a simple vector, and if the number and the types of the elements are know? i woldn't know what size to set for truncation. On VS 2010, a sizeof(vector<int>) is 40; I suppose I have to add it sizeof(int)*nbElements to get the full size. 1. I don't know if this assumption is valid 2. If so, I don't know if it's still valid when using Boost.Interprocess (I suppose it isn't) 3. I don't know how to generalize that to a vector of vectors in order what should be the truncation size. Thanks for help. PS : I also read http://www.boost.org/doc/libs/1_52_0/doc/html/interprocess/managed_memory_s…, but I havn't been enlightened about these points.

4 14

Interprocess : offset_ptr used as array
by Oodini 16 May '13

16 May '13

Hello, I have an offset_ptr named tmpData defined like this : offset_ptr<const byte>* someVariable; indextype index; ... offset_ptr<const byte> tmpData = ((offset_ptr<const byte>*)someVariable)[index]; data points actually to the beginning of an array. The original code, without shared memory was : byte* data= ((byte**)m_someVariable)[index]; The problem is that I have to translate that : indextype* data = (indextype*)&(tmpData[1]); The adress of the second element is taken though the & adress. But when tmpData is an offset_ptr, the [] operator returns a reference, and I can apply the & operator. Is there any solution that copying the pointer to temporary variable, and increment this one ? Thanks.

1 0