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

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

tuple problems
by programmer blabla 14 Jan '04

14 Jan '04

Dear Booster's 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? 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? Btw searching lists.boost.org is very hard, because google is out of sync with the list. With kind regards, Me.

1 0

int64_t in boost/cstdint.h with GCC -std=c++98 or -ansi
by Aaron W. LaFramboise 14 Jan '04

14 Jan '04

Hi, I noticed that boost/cstdint.h does not define int64_t when GCC is in strict mode (-ansi or -std=c++98) on MinGW, Cygwin, and probably other major platforms. This seems to be because boost/cstdint.h refuses to define the type if ULLONG_MAX (or ULONG_LONG_MAX or ULONGLONG_MAX) is not defined, even if BOOST_HAS_LONG_LONG is defined. GCC, quite correctly, will not define these macros in strict mode. However, the library still probably should be defining this type. Ideally, there would be a GCC-specific work-around to prevent warnings on the use of long long in strict mode. In particular, I am using the date_time library which depends on int64_t. It would be very unfortunate and a little silly if I could not use date_time in GCC's strict mode. How would it be best to fix this? Aaron

2 2

d:\sdk\boost\boost\mpl\list\aux_\iterator.hpp(31) : error C2039: 'item' : is not a member of 'Node'
by pixellent＠stodge.net 14 Jan '04

14 Jan '04

Hi, I'm trying to use Luabind in my application, and Luabind uses BOOST. I'm getting the following error: d:\sdk\boost\boost\mpl\list\aux_\iterator.hpp(31) : error C2039: 'item' : is not a member of 'Node' d:\development\ogre\ogrenew\ogremain\include\ogrenode.h(50) : see declaration of 'Node' d:\sdk\boost\boost\mpl\list\aux_\iterator.hpp(33) : see reference to class template instantiation 'boost::mpl::list_iterator<Node>' being compiled d:\sdk\boost\boost\mpl\list\aux_\iterator.hpp(31) : error C2146: syntax error : missing ';' before identifier 'type' d:\sdk\boost\boost\mpl\list\aux_\iterator.hpp(33) : see reference to class template instantiation 'boost::mpl::list_iterator<Node>' being compiled d:\sdk\boost\boost\mpl\list\aux_\iterator.hpp(31) : error C2208: 'struct boost::type' : no members defined using this type d:\sdk\boost\boost\mpl\list\aux_\iterator.hpp(33) : see reference to class template instantiation 'boost::mpl::list_iterator<Node>' being compiled d:\sdk\boost\boost\mpl\list\aux_\iterator.hpp(32) : error C2039: 'next' : is not a member of 'Node' d:\development\ogre\ogrenew\ogremain\include\ogrenode.h(50) : see declaration of 'Node' d:\sdk\boost\boost\mpl\list\aux_\iterator.hpp(33) : see reference to class template instantiation 'boost::mpl::list_iterator<Node>' being compiled d:\sdk\boost\boost\mpl\list\aux_\iterator.hpp(32) : error C2146: syntax error : missing ',' before identifier 'next' d:\sdk\boost\boost\mpl\list\aux_\iterator.hpp(33) : see reference to class template instantiation 'boost::mpl::list_iterator<Node>' being compiled d:\sdk\boost\boost\mpl\list\aux_\iterator.hpp(32) : error C2065: 'next' : undeclared identifier d:\sdk\boost\boost\mpl\list\aux_\iterator.hpp(33) : see reference to class template instantiation 'boost::mpl::list_iterator<Node>' being compiled d:\development\ogre\ogrenew\plugins\ogrelua\luabind\luabind\detail\primitives.hpp(91) : fatal error C1506: unrecoverable block scoping error Error executing cl.exe. Nature.obj - 7 error(s), 0 warning(s) Is this a bug in BOOST or am I doing something silly? Thanks Mike

1 0

Seeking permission to include Boost with book CD
by Matthew Wilson 13 Jan '04

13 Jan '04

Hi chaps / chappettes I am interested in including the latest released copy of Boost with the CD for my forthcoming book, Imperfect C++, to be published by Addison-Wesley this year. I have two questions: 1. Is Boost redistributable in this form? If so, from whom do I get written permission to do so? 2. What is the likely latest version of Boost that will be available around the end of Feb / beginning of March? Thanks in advance -- Matthew Wilson Director, Synesis Software (www.synesis.com.au) STLSoft moderator (http://www.stlsoft.org) Contributing editor, C/C++ Users Journal (www.synesis.com.au/articles.html#columns) -----------------------------------------------------

2 2

re: API documentation ?
by Vikram Anantha 13 Jan '04

13 Jan '04

Is there any place where I can look at the complete API documentation for the boost libraries. Something generated by doxygen or doc++ or something like that. Thanks Vik

2 1