January 2004 - Boost-users - lists.preview.boost.org

RE: [Boost-users] tuple problems
by Ben Hutchings 17 Jan '04

17 Jan '04

programmer blabla <programmer_blabla(a)hotmail.com> wrote: > Dear Booster's Booster's what? > I have a problem compiling a tuple containing an enum in a > namespace. For example: > > namespace TestTuple > { > enum ETestTuple > { > eTestTuple0 = 0, > eTestTuple1 = 1 > }; > } > > void Testtuple5() > { > typedef boost::tuple<TestTuple::ETestTuple> Tuple; > Tuple tpl; > } > > does not compile. I am using MSVC 2003.net (7.1 version), > just native code. Is this a M$ problem or is tuple broken? Looks like a compiler bug to me. > Also I get problem with > the combination > of boost::bind and tuple, for example: > > int i; > boost::tuple<int> data; > > i = data.get<0>(); //ok > i = boost::tuples::get<0>(data); //ok > i = boost::bind(&boost::tuples::get<0>, _1)(data); //error > > Has anyone a clue? boost::tuples::get has three template parameters - index, head type and tail type. Usually the type arguments can be deduced, but since you're passing the address to another template function you need to specify all three arguments. > Btw searching lists.boost.org is very hard, because google > is out of sync with the list. Try <http://news.gmane.org/index.php?prefix=gmane.comp.lib.boost> instead.

2 1

[BGL] Custom edge-removing visitor.
by Kili2x＠netscape.net 17 Jan '04

17 Jan '04

Hello, all. I'm working on an algorithm that takes a graph and two path iterators (begin and past-the-end) as input, and removes "cross edges" (is that the right term in this case?) from the graph until the path described by the iterators is the only simple path between those two vertices. The implementation I have so far consists of a function template and three visitor class templates built on top of the boost::depth_first_visit function template. I'm still improving on it (too many out-edges from the path itself are being removed), but for now the function works as it should. Now, the graph reference passed to the operator() member function in the boost::base_visitor class template (from which I inherited) must be a constant reference, so I can't pass it to the boost::remove_edge function. Furthermore, looking at the source code included in <boost/graph/depth_first_search.hpp> (both 1.30.2 and CVS versions), the edge passed to vis.forward_or_cross_edge cannot be removed without invalidating the out_edge_iterator used in the depth-first traversal. I've worked around the first issue by keeping a non-constant reference to the graph as a member of my custom visitor class and passing that reference to boost::remove_edge, and I'm using the boost::on_finish_vertex event filter instead of boost::on_forward_or_cross_edge (adjusting the code as appropritate) in order to overcome the second issue. The workarounds look rather kludgey in code, so removing the const modifier and/or adding an extra iterator to store the current edge while incrementing the traversing iterator might enable me to come up with a more elegant solution. Would such changes to <boost/graph/depth_first_search.hpp> break compatibility with older code? Or are there other particular issues in play? BTW, Is this the appropriate list for seeking such help? TIA! Cromwell Enage __________________________________________________________________ New! Unlimited Netscape Internet Service. Only $9.95 a month -- Sign up today at http://isp.netscape.com/register Act now to get a personalized email address! Netscape. Just the Net You Need.

1 0

question about mpl::list
by Istvan Buki 16 Jan '04

16 Jan '04

Hi all, I'm using the mpl::list template in one of my aplications but I just hit what seems to be a limit. list compile fine with 10 parameters but not with 11. I saw in the documentation that there are several header files like : #include "boost/mpl/list/list50.hpp" Are they meant to be used in cases like mine and if it is the case could someone point me to a short example ? Thanks for your help, Istvan

2 2

interest in a glob_iterator? (directory_iterator with regex)
by Rich Johnson 16 Jan '04

16 Jan '04

Folks-- Is there general interest in a "globbing" iterator? If so I've got one I'm willing to re-package for submission. (Or at least discuss with folks for improvement) If you're interested, read on: glob_iterator aggregates a directory_iterator and regex to provide shell-style "*", "?", "{....}", "[....]" and "[^...]" wildcarding. boost components used: filesystem::directory_iterator filesystem::path filter_iterator reg_expression c_regex_traits Usage example: //...do something to all .cpp and .c files glob_iterator start( "*.{c,cpp}" ); glob_iterator end; while( start != end ){ std::string filename( start->leaf() ); //...do something with/to filename ++start; }

8 11

Boost.Python Feature Request
by Jenna Louis 16 Jan '04

16 Jan '04

Are there any patches or plans to add the following features to Boost.Python/Pyste? 1. A major usability problem of Boost generated proxies is that IDEs are not able to show any details about the expected parameters. Since Boost already knows everything about the function then it makes sense to at least make some information known about it via a docstring and then these tools can at least show that. 2. Another feature would append a docstring with the descriptive text to the specified Class, Function, or Method. 3. I often generate modules that require a very specific order of initialization for the proxies it contains. I often have to rearrange the _main.cpp's BOOST_PYTHON_MODULE code as depicted below to get my module to work. This: BOOST_PYTHON_MODULE(Sim) { Export_SzCoordPos(); Export_SzCoordAng(); Export_SzCoord(); Export_SzMath(); Export_SzPolyUtils() } To This: BOOST_PYTHON_MODULE(Sim) { Export_SzCoord(); Export_SzMath(); Export_SzPolyUtils() Export_SzCoordPos(); Export_SzCoordAng(); } It would be nice if this list would be generated in the order in which the pyste files are specified on the command line. 4. I would like a way to add pickling code to my class from pyste. I realize its not practical to generate this code automatically but I would at least like to be able to inline this in my pyste file. I can already insert code into the proxy with the declaration_code() feature to handle the pickle struct but I also need to be able to add a pickle method i.e.(".def_pickle(world_pickle_suite())" to the class definition I specify. Thanks, Jenna --------------------------------- Do you Yahoo!? Yahoo! Hotjobs: Enter the "Signing Bonus" Sweepstakes

2 1

RE: [Boost-users] Re: interest in a glob_iterator? (directory_ite rator with regex)
by Jim.Hyslop 15 Jan '04

15 Jan '04

Angus Leeming [mailto:angus.leeming@btopenworld.com] wrote: > Jeff Flinn wrote: > > And a more descriptive, less jargon-ish name. > > Why? Rich is proposing something that would iterate over the files > returned by the unix 'glob' function. As much as you might hate to admit it, you must recognize that the world does not revolve around UNIX (and before anyone claims otherwise, nor does the world revolve around Windows). To my knowledge, "glob" is not a standard computer term, nor a standard C++ term, it is a standard UNIX term. Now, I could be wrong, but my understanding of the ISO standardization process is make everything platform-independent (and keep in mind that one of the original goals of Boost was to be included in the Standard). The *concept* of the "glob" is pretty much platform independent, but the term is not, so it should have (to paraphrase Jeff) a more descriptive, less platform-specific name. -- Jim Hyslop Senior Software Designer Leitch Technology International Inc. (<http://www.leitch.com/>) Columnist, C/C++ Users Journal (<http://www.cuj.com/experts>)

3 2

boost::scoped_ptr and std::auto_ptr interoperability
by Sam Partington 15 Jan '04

15 Jan '04

Hi all, Is there any reason why boost::scoped_ptr::reset doesn't provide a std::auto_ptr overload? Take the following code : class A {}; std::auto_ptr<A> factory(); // ... boost::scoped_ptr<A> a; a.reset(factory().release()); wheras a.reset(factory()); would be neater. (Trivial difference I know, but it does add to readability) Sam --LongSig ****************** This e-mail has been sent from Imagination Technologies Limited. PowerVR, Metagence, Ensigma and PURE Digital are divisions of Imagination Technologies Limited. The information contained in this e-mail, including any attachment, is confidential and may be legally privileged. It is intended solely for the addressee(s) and access to this e-mail by anyone else is unauthorised. If you are not the intended recipient, any disclosure, copying or distribution or use of the information contained in this e-mail, is prohibited and may be unlawful. If you have received this e-mail in error, please notify the sender by return e-mail and then delete it from your system. Internet communications cannot be guaranteed to be secure, error or virus-free. The sender does not accept liability for any errors or omissions which arise as a result. Any views expressed in this message are those of the author, except where the author specifies and, with authority, states them to be the views of Imagination Technologies Ltd.

1 0

boost on Mac OS X 10.3 (Panther)
by Vineet Mehta 15 Jan '04

15 Jan '04

Has anyone been able to successfully build and install boost on Mac OS X 10.3 (panther). I am using gcc 3.3, the latest xcode. The python variables were configured to: export PYTHON_ROOT=/System/Library/Frameworks/Python.framework/Versions/2.3 export PYTHON_VERSION=2.3 I have bjam installed. I used bjam "-sTOOLS=darwin" in the boost root directory. However at the yields a number of warning and error messages, and ends with : ...failed updating 70 targets... .. .. Any suggestions ? -Vineet

8 9

Problems with Reverse Cuthill-Mckee?
by Kris Vorwerk 14 Jan '04

14 Jan '04

Hello, I'm a fairly new Boost user, so please forgive me if I'm completely out of whack here. I'm presently using Boost's reverse Cuthill-Mckee ordering to reduce the bandwidth of some very large symmetric, positive definite sparse matrices that I'm investigating. Generally, the cuthill_mckee_ordering() has worked quite well, although a short while ago, I began noticing that it occasionally gave an invalid ordering (i.e., the permutations spit out by the algorithm would have multiple vertexes all the same, which, when used to permute my matrices, would destroy my problem). It took me a while to figure out what was going on, but I've finally distilled a small enough example that exhibits the problem. Consider the following matrix: Matrix: 6x6 [ [3.66667,0,-0.5,0,0,0], [0,2.66667,0,-0.222222,0,0], [-0.5,0,1.66667,0,0,-0.222222], [0,-0.222222,0,3.33333,-2.22222,0], [0,0,0,-2.22222,4.66667,0], [0,0,-0.222222,0,0,4.66667] ] Using fairly straightforward code, I loop through the non-zeros, and add edges to an undirected Boost graph (basically the same as the cuthill_mckee_ordering.cpp example in the docs). Here is the graph that gets created for the above problem, in Graphviz format: graph G { 0; 1; 2; 3; 4; 5; 0--0 ; 0--2 ; 1--1 ; 1--3 ; 2--2 ; 2--5 ; 3--3 ; 3--4 ; 4--4 ; 5--5 ; } And, if I run this example using the cuthill_mckee_ordering technique (or if I modify cuthill_mckee_ordering.cpp to use these vertices and edges), I get the following permutation (using the pseudo-peripheral heuristic): Inv_perm: 0 0 0 0 2 5 You can see, from the Graphviz output, that this graph has two disconnected sets of vertices, which is where I think the problem lies. It may be a problem in my understanding of how to setup the Boost graph, or perhaps a problem in the reordering technique. You'll notice that, in my example, I've eliminated the duplicate edges (such as 2-5 being the same as 5-2) -- I've tried adding them in, too, but it hasn't changed the result. I've noticed that, if one of my matrices produces a graph in which there are no disconnected sets of vertices, the RCM ordering works fine. It's only when the graph is disconnected in this fashion that the ordering fails. I've also noticed that, when using this kind of disconnected graph, if I sometimes select a random starting vertex (instead of the pseudo-peripheral pair heuristic), the cuthill_mckee_ordering segfaults. I've attached some sample code below that seems to exhibit the problem. I'm certainly open to the possibility that it's a problem in my understanding of how to use the library. Also, my Boost version is: 1.30.2 (Redhat RPM) Any thoughts would be appreciated. regards, Kris ============================================================== (borrowed and slightly modified from cuthill_mckee_ordering.cpp) #include <boost/config.hpp> #include <vector> #include <iostream> #include <boost/graph/adjacency_list.hpp> #include <boost/graph/cuthill_mckee_ordering.hpp> #include <boost/graph/properties.hpp> #include <boost/graph/bandwidth.hpp> int main( int, char *[] ) { using namespace boost; using namespace std; typedef adjacency_list < vecS, vecS, undirectedS, property < vertex_color_t, default_color_type, property < vertex_degree_t, int > > > Graph; typedef graph_traits < Graph >::vertex_descriptor Vertex; typedef graph_traits < Graph >::vertices_size_type size_type; typedef std::pair < std::size_t, std::size_t > Pair; //graph G {0;1;2;3;4;5;0--0 ;0--2 ;1--1 ;1--3 ;2--2 ;2--5 ;3--3 ;3--4 ;4--4 ;5--5 ;} Pair edges[] = { Pair( 0, 0 ), Pair( 0, 2 ), Pair( 1, 1 ), Pair( 1, 3 ), Pair( 2, 2 ), Pair( 2, 5 ), Pair( 3, 3 ), Pair( 3, 4 ), Pair( 4, 4 ), Pair( 5, 5 )}; Graph G( 6 ); for( int i = 0; i < 10; ++i ) { std::cout << edges[i].first << ", " << edges[i].second << std::endl; add_edge( edges[i].first, edges[i].second, G ); } graph_traits < Graph >::vertex_iterator ui, ui_end; property_map < Graph, vertex_degree_t >::type deg = get( vertex_degree, G ); for( boost::tie( ui, ui_end ) = vertices( G ); ui != ui_end; ++ui ) { deg[*ui] = degree( *ui, G ); } property_map < Graph, vertex_index_t >::type index_map = get( vertex_index, G ); std::cout << "original bandwidth: " << bandwidth( G ) << std::endl; std::vector < Vertex > inv_perm( num_vertices( G ) ); std::vector < size_type > perm( num_vertices( G ) ); // reverse cuthill_mckee_ordering cuthill_mckee_ordering( G, inv_perm.rbegin(), get( vertex_color, G ), make_degree_map( G ) ); cout << "Reverse Cuthill-McKee ordering:" << endl; cout << " "; for( std::vector < Vertex >::const_iterator i = inv_perm.begin(); i != inv_perm.end(); ++i ) { cout << index_map[*i] << " "; } cout << endl; for( size_type c = 0; c != inv_perm.size(); ++c ) { perm[index_map[inv_perm[c]]] = c; } std::cout << " bandwidth: " << bandwidth( G, make_iterator_property_map( &perm[0], index_map, perm[0] ) ) << std::endl; return 0; }

2 3

Re:[Boost-users] interest in a glob_iterator? (directory_ite
by rlyon＠comgroup.net.au 14 Jan '04

14 Jan '04

I can use it. Hopefully you will use regex++ from boost to handle the regular expressions. Regards ... ____________________Reply Separator____________________ Subject: [Boost-users] interest in a glob_iterator? (directory_iterat Author: Rich Johnson <rjohnson(a)dogstar-interactive.com> Date: 13/01/2004 6:45 PM Folks-- Is there general interest in a "globbing" iterator? If so I've got one I'm willing to re-package for submission. (Or at least discuss with folks for improvement)

1 0