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

Re: [Boost-users] Boost-users Digest, Vol 2767, Issue 2
by Marat Abrarov 29 Jun '11

29 Jun '11

> I am grateful for the link, Marat, although I can't do exactly what you are doing. We are constrained > as to what tools we can use with our embedded target's Linux. Oh, I missed this. > I've got to use the proper gcc, gmake, > and libs. It's going to be painful enough to get the target's boost libraries upgraded because some of > them will cause API breakages. Boost.Asio is header only library. I often upgrade Boost with most recent Boost.Asio version by overwriting files. Regards, Marat Abrarov.

2 1

BGL Traveling Salesman Approximation
by David Doria 29 Jun '11

29 Jun '11

Using the example here: http://www.boost.org/doc/libs/1_44_0/libs/graph/test/metric_tsp_approx.cpp I constructed this path: http://homepages.rpi.edu/~doriad/Upload/contour.png on these points: http://homepages.rpi.edu/~doriad/Upload/points.png Is this the best I can hope to do with this approximate algorithm? (There is "clearly" (to a human) a much shorter path without all of the overlapping edges on the lower part of the left side of the rectangle). The graph.txt file I used is here: http://homepages.rpi.edu/~doriad/Upload/graph.txt and the solution it gave was: 0 1 2 3 5 7 9 11 13 15 16 18 20 21 23 25 26 28 30 32 35 37 40 39 38 36 34 33 31 29 27 24 22 4 6 8 10 12 14 17 19 0 Is there anyway I can "fix" this solution? Thanks, David

2 1

filesystem3::rename error codes
by Andreas Haferburg 29 Jun '11

29 Jun '11

Greetings, I was wondering how to handle errors of filesystem3::rename(). I'd like to use the system::error_code& overload, but I can't seem to find out how to use it. The documentation at [1] points to POSIX rename(), so I'm guessing the error_code somehow corresponds to the POSIX rename() return value. I'm not a native speaker, but I find the documentation pretty vague here. The tutorial section [2] doesn't help either. For example if the target file doesn't exist, the error_code.value() is 3. But I'm not supposed to use the constant 3 in the code, am I? Are there any constants defined that correspond to these error code values? What's the proper way to do the following: TEST( BoostFilesystem3Test, testRenameFails ) { boost::system::error_code ec; boost::filesystem3::rename("c:/does/not/exist", "asdf", ec); ASSERT_EQ(3, ec.value()); } I'm going to use QFile for now, but only because their documentation is (slightly) better. I would like to use boost for as much as possible, so an explanation or pointers to better documentation would be much appreciated. Thank you. Best regards Andreas P.S. I accidentally sent this mail to the boost@ list before, sorry about that. [1] http://www.boost.org/doc/libs/1_46_1/libs/filesystem/v3/doc/reference.html#… [2] http://www.boost.org/doc/libs/1_46_1/libs/filesystem/v3/doc/tutorial.html#E…

1 0

Using Boost::Bind
by David McCallum 29 Jun '11

29 Jun '11

I'm having some difficulty passing parameters to boost::bind. The approach taken by ASIO is to pass a placeholder to bind, like so: hostResolver->async_resolve(*nameQuery, boost::bind(&DClient::resolveHandler, this, boost::asio::placeholders::error, boost::asio::placeholders::iterator)); This works fine if I want to use an error or iterator as a parameter. But what if I want to use a primitive variable, or a pointer to one of my classes? Do I need to make a placeholder for each type of parameter I want to use? I looked in ASIO's placeholders.hpp file, and I must confess I can't make sense of it at all! Any help here is appreciated.

3 3

threading/io service query
by Sean Farrow 28 Jun '11

28 Jun '11

Hi Fokes: I'm currently working on a piece of software that is using or going to use boost. There are two classes that use the boost threading components: The first MidiWidget allows the opening of midi input/output ports, I plan to associate two threads, one to receive data and one to output the data. The second is where the problem could occur, MidiSurface simulates a midi control surface, which processes midi messages. Given that there could be multiple surfaces in the program-determined by connections to the pc, and that each surface may need to process a different set of messages which are defined in an xml file, what do people consider is the best approach to firing up a thread when ever a surface is created. I'd ideally like each surfaces thread be able to examine the data contained within it's owning surface without copying the whole class. Is it better to have the thread function as a virtual function in the surface class or a separate function that takes the surface, or a reference to the surface as a parameter. Also I've got a thread safe queue, but was wondering whether there is now one included with boost-or tr1. I'm using boost 1.46.1 and visual studio 2010 sp1 with the latest windows sdk. Any help/thoughts appreciateds. Regards Sean.

1 0

BGL: Copying graphs and edge_descriptors confusion
by David Doria 28 Jun '11

28 Jun '11

I have a graph, 'inputGraph' with a bool in its bundled edge properties. If I copy that graph with: Graph g = inputGraph; there seems to be some odd behavior with the edge_descriptors. If you take a look at this example: http://programmingexamples.net/index.php?title=CPP/Boost/BGL/CopyAGraph I set one of the edges to visible=0 and then output all of the edge visibilities using the iterator from boost::edges(g) - this works properly, the edge that I set to visible=0 indeeds shows up as such. However, if I instead use the iterator from boost::edges(inputGraph) , all of the edges appear as visible! Shouldn't the edge_descriptors be identical since the graph g is an exact copy of inputGraph? Can anyone explain why the last two loops in that example do not produce the same output? Thanks, David

3 5

read_graphviz with a simple graph
by David Doria 28 Jun '11

28 Jun '11

If I have a graph with no vertex properties or edge properties: adjacency_list<vecS, vecS, undirectedS> can I read a .dot file into it with read_graphviz? I tried: dynamic_properties dp; Graph g; read_graphviz(fileinput, g, dp); but it throws an exception "Property not found: node_id". I tried to add property_map<Graph, vertex_name_t>::type name = get(vertex_name, g); dp.property("node_id",name); but I get a compiler error "No match for operator<<" Any suggestions? Thanks, David

2 14

Re: [Boost-users] boost::asio and C++0x
by Marat Abrarov 28 Jun '11

28 Jun '11

> It appears that gcc up to 4.5.2 did work; the bug was related to bleeding-edge gcc. > > I am still looking for solutions to this issue. I have successfully built my app with -std=c++0x and > boost::thread, but boost::asio gives me errors. At http://sourceforge.net/projects/asio-samples/ I use qmake to build projects under Linux. I build asio-samples with Boost 1.46.0 + Boost.Asio 1.5.3 + gcc 4.6.0. Feel free to watch at *.pro files describing qmake build dependencies at /build/qmake. Regards, Marat Abrarov.

2 1

boost::asio and C++0x
by Wilde, Donald S 28 Jun '11

28 Jun '11

Hi, All - I am having some challenges mixing boost and c++0x. My configuration: Ubuntu 10.04, gcc 4.4.3, boost 1.46.1 I've updated the libraries and headers from Ubuntu's stock boost 1.40 level, and I've tried bjam with and without 'toolset=gcc cxxflags=-std=c++0x' Either way, my Eclipse reports several errors in the external libraries: Undefined references to boost::system::generic_category() Undefined references to boost::system::system_category() (both from system and asio error headers) Undefined references to pthread_key_create and pthread_key_delete from asio/detail headers { apologies for not giving specific error details but it'd have to come from a screen shot and that's not appreciated on a mailing list } I do have access to gcc 4.5.1, but it's a custom one for our embedded Atom-based SoC. That is the eventual target, but I haven't seen how to use a specific g++ binary yet. I did see a bug tracker post in one of my google searches but I haven't been able to get back to it. I understand that this is a known issue and a moving target but I would appreciate it if someone would summarize the current state for me. Thanks in advance! -- Don Wilde

3 6

Multi_array and rotated views
by petros 28 Jun '11

28 Jun '11

Hi, I am coming back to the list with a question on multi_array views. Some time ago, I asked if I could have a view that would offer a “rotated” multi_array. That is, if I have a 3D MA (=multi_array) of dimensions (shape) {L,M,N}), I would like to obtain a reference to it which would look like {M,N,L} i.e. cyclically rotated (I would actually wish to be able to perform this rotation many times). The operation should be performed at compile time (no need/wish to do it at run time and pay the possible overhead in speed). A couple of friends suggested that I used a storage structure something like this (files are attached for convenience of inspection). namespace boost { class cyclic_storage_order { typedef detail::multi_array::size_type size_type; public: cyclic_storage_order( const std::size_t shift ) :shift_(shift) {} template <std::size_t NumDims> operator general_storage_order<NumDims>() const { boost::array<size_type,NumDims> ordering; boost::array<bool,NumDims> ascending; for (size_type i=0; i != NumDims; ++i) { ordering[i] = (i+shift_)%NumDims; ascending[i] = true; } return general_storage_order<NumDims>(ordering.begin(), ascending.begin()); } private: std::size_t shift_; }; } Now, I am trying to use this is the following example (again the file is attached): #define _SCL_SECURE_NO_WARNINGS #include <iostream> using std::cout; using std::endl; #include <boost/multi_array.hpp> using boost::multi_array; using boost::multi_array_ref; using boost::const_multi_array_ref; #include "MultiArrayRotation.h" int main(){ const unsigned short dimensionality = 3; typedef multi_array< int, dimensionality > IntArray; typedef IntArray::size_type size_type; typedef const_multi_array_ref< int, dimensionality > IntArrayCRef; boost::array<size_type,dimensionality> sizes = { { 2,3,4 } }; IntArray a( sizes ); boost::cyclic_storage_order cyclic_storage(1); boost::array<size_type,dimensionality> sizesRotated = { { 3, 4, 2 } }; IntArrayCRef rotated( a.data(), sizesRotated, cyclic_storage ); //initialize: for ( size_t i = 0 ; i != sizes[0]; ++i ) for ( size_t j = 0 ; j != sizes[1]; ++j ) for ( size_t k = 0 ; k != sizes[2]; ++k ) a[i][j][k] = (i+1) + 10 * (j+1) + 100*(k+1); typedef IntArray::const_iterator iterator3; typedef boost::subarray_gen<IntArray,2>::type::const_iterator iterator2; typedef boost::subarray_gen<IntArray,1>::type::const_iterator iterator1; for ( iterator3 it3 = a.begin(); it3 != a.end(); ++ it3 ) { cout << "in first iterator:"; boost::detail::multi_array::const_sub_array<int,2> & d2 = *it3; cout << (d2.shape())[0] << "x" << (d2.shape())[1] << endl; for ( iterator2 it2 = it3->begin(); it2 != it3->end(); ++ it2 ) { // cout << "\t" << it2->shape()[0] << endl; for ( iterator1 it1 = it2->begin(); it1 != it2->end(); ++ it1 ) cout << "\t\t" << (*it1) << endl; } } //for a rotated system: typedef IntArrayCRef::const_iterator iterator3CRef; typedef boost::subarray_gen<IntArrayCRef,2>::type::const_iterator iterator2CRef; typedef boost::subarray_gen<IntArrayCRef,1>::type::const_iterator iterator1CRef; cout << endl << "In Rotated System" << endl; for ( iterator3CRef it3 = rotated.begin(); it3 != rotated.end(); ++ it3 ) { cout << "in first iterator:"; //boost::const_multi_array_ref::const_sub_array<int,2> & d2 = *it3; cout << (it3->shape())[0] << "x" << (it3->shape())[1] << endl; for ( iterator2CRef it2 = it3->begin(); it2 != it3->end(); ++ it2 ) { // cout << "\t" << it2->shape()[0] << endl; for ( iterator1CRef it1 = it2->begin(); it1 != it2->end(); ++ it1 ) cout << "\t\t" << (*it1) << endl; } } return 1; } This gives me undesired results, though. What I have done here, is define a 2x3x4 main array with entries like abc (i.e. 3 digits). >From the value of abc I can infer the coordinates of the element in the main array. Following the cascade of iterators to sub_arrays, I see that, in the end, I obtain for each value of the first coordinate, the corresponding slices. Then I do a rotation (once) and descend down the iterators. I would expect now 3 slices (instead of 2, before), organized so that each one has the same middle digit (in the abc “representation”. This, however, is not happenning! I attach the code which I have built successfully w/msvc2010 on win7. I use boost1.45. A friend (Larry Evans) , who had offered a lot of help in the past and whom I contacted for suggestion let me know that, the program did not compile in his environment (gcc 4.6.0 and boost v1.46). Any suggestion, help will be greatly appreciated. Kind Regards, Petros

2 3