February 2005 - Boost-users - lists.preview.boost.org

Graph Library Question
by Todd A. Gibson 18 Mar '05

18 Mar '05

Hello, I have a code base which uses adjacency_list as the underlying graph structure. I tried converting from adjacency_list to adjacency_matrix and discovered that adjacency_matrix does not implement the in_edges() non-member function (nor the corresponding iterators). What is the design rationale for including only out_edges() in adjacency_matrix but not in_edges()? Is there a recommended alternative for traversing in_edges for an adjacency_matrix? Thanks, -TAG

2 1

Bug in boost/graph/transitive_closure.hpp
by Todd A. Gibson 14 Mar '05

14 Mar '05

Bug: The header file boost/graph/transitive_closure.hpp contains an unscoped call to not1(). This function is defined in both the std namespace and the boost namespace (via boost/functional.hpp). If boost/functional.hpp is included either directly or indirectly (e.g., via boost/multi_array.hpp) Then any attempted use of transitive_closure() will not compile due to the ambiguity. Solution: Scope the call to not1 in boost/graph/transitive_closure.hpp: std::not1 Note: This bug can be easily be reproduced by including boost/functional.hpp (or boost/multi_array.hpp) To the example program transitive_closure.cpp . Thanks, -TAG Todd A. Gibson

3 3

BGL: EdgeList template parameter specification for undirected graphs
by Anatoly Petkevich 14 Mar '05

14 Mar '05

Hi! I am trying to use the BGL to create sparse undirected graph using boost::setS as template argument for EdgeList parameter: typedef adjacency_list<vecS, vecS, undirectedS, no_property, no_property, no_property, setS> GraphType; Impementation of boost::add_edge() for undirected graphs calls g.m_edges.push_back(e), and we get compilation error here since push_back() is not a member of associative container std::set. There is no such a problem for directed graphs (directedS). According to the BGL documentation there are no restrictions on EdgeList parameter for undirected graphs. Is this an undocumented feature or inconsistent implementation or library misuse? Regards, A.Petkevich

3 3

fruchterman_reingold doesn't like small graphs
by Jamie Wilkinson 14 Mar '05

14 Mar '05

I'm writing an app in which the user incrementally adds vertices to the graph; of course it starts empty, and after each addition the graph is laid out with the fruchterman-reingold function. The algorithm copes well with empty graphs, but once I add a single vertex, the program crashes: the crash is always in the same place, deep in std::list but called from boost/graph/fruchterman_reingold.hpp:116 : buckets[row * columns + column].push_back(*v); The push_back function calls end() to get the last element of the list, and fails. A bit of gdb turned up the following result: columns and rows were both being set to 0 and so buckets was also of size 0. Lines 111:112 also end up setting column and row to 0, and then the next pair of lines were setting them to, on my machine, 4G. The element of buckets being accessed, then, was ridiculously out of bounds. The value width/two_k, once cast to the integral size_t, was always zero. I don't think this is restricted to having a single vertex in the graph, but that case seems to exacerbate the bug. So as a workaround, I've added the following two lines to make sure that the variables columns and rows are at least 1: Index: boost/graph/fruchterman_reingold.hpp =================================================================== RCS file: /cvsroot/boost/boost/boost/graph/fruchterman_reingold.hpp,v retrieving revision 1.5 diff -u -r1.5 fruchterman_reingold.hpp --- boost/graph/fruchterman_reingold.hpp 29 Dec 2004 16:53:05 -0000 1.5 +++ boost/graph/fruchterman_reingold.hpp 13 Jan 2005 13:09:00 -0000 @@ -105,6 +105,8 @@ std::size_t columns = std::size_t(width / two_k); std::size_t rows = std::size_t(height / two_k); + if (columns == 0) columns = 1; + if (rows == 0) rows = 1; buckets_t buckets(rows * columns); vertex_iterator v, v_end; for (tie(v, v_end) = vertices(g); v != v_end; ++v) { If you wish to reproduce this bug, you can do so with the example program libs/graph/example/fr_layout.cpp : g++ -Iboost boost/libs/graph/example/fr_layout.cpp echo "A B" | ./a.out 100 100 My results without the patched function: willow% echo "A B" | ./a.out 100 100 0% 10 20 30 40 50 60 70 80 90 100% |----|----|----|----|----|----|----|----|----|----| *zsh: done echo "A B" | zsh: segmentation fault ./a.out 100 100 and patched: willow% echo "A B" | ./a.out 100 100 0% 10 20 30 40 50 60 70 80 90 100% |----|----|----|----|----|----|----|----|----|----| *************************************************** A -41.0322 -29.471 B 1.16969 27.1354

2 1

indirect_iterator of reverse_iterator
by Thomas Matelich 08 Mar '05

08 Mar '05

Hello, I'm a little behind the times and am trying to port my old iterator_adapters code to use the new library (1.32). We have a class which is little more than a wrapper around a vector<shared_ptr<T> >, and I am having a heck of a time with setting its typedef for reverse_iterator and const_reverse_iterator. I just can't get const_reverse_iterator to do assignment and comparison with reverse_iterator. I've tried to make indirect_iterators out of the native reverse iterators, using boost::reverse_iterators of indirect_iterators, and a number of more complicated combinations of the 3. No matter which I try I get errors about the const_reverse_iterator not playing nicely with the reverse_iterator. I think I've read everything on the list archives about indirect_iterator, but perhaps I missed something. I'm going to try my code on g++ (currently using VC 7.1) and then work on a complete example, but I thought I'd throw this out and see if someone can quickly answer it. Thanks, Tom

2 6

[regex] Is there any function to know where the input fails to match a regex?
by jordi 07 Mar '05

07 Mar '05

Hello, Maybe my question is a silly question but I have a recurrent problem and maybe there is an easy solution: I need to create a lot of regex expressions to match several Unix command outputs. Usually, my firsts regex definitions are wrong and they don't match, so when the match fails I would like to know where the "problem" is (more or less) in order to change the regular expression. For example: Input text: "123 hello abc" My first wrong Regex: "(\d)+\s+(\w)+\s+(\d)+" If I try to match this input against the regex using regex_search with the boost::match_continuous flag I only have the false value returned by the function but it doesn't help me to know that the "123 hello" matched the regex because the problem starts at "abc" trying to match "(\d)+". I can simulate what I need checking subexpressions adding every iteration a new "sub-regex" and stopping when the match fails (I do it manually, changing the regex). With my undefined funtion I would obtain: - Input text:"123 hello abc" - Latest rigth Regular expression: "(\d)+\s+(\w)+\s+" Matches: "123 hello " - First wrong regex: "(\d)+\s+(\w)+\s+(\d)+" does not match. -> So, "(\d)+" does not match "abc" I understand that usually this is not as easy as I write in my simple example because of the power of regular expressions but I think a function like that would improve a lot the error reporting capabilities. Is there any way to do what I need or I must to write my own check function? Best regards, Jordi

4 7

Borland compiler error E2437
by Roman Morokutti 07 Mar '05

07 Mar '05

Hi, I am not able to compile function_base.hpp at lines 595(55), 607(55), 616(55), 625(55). Error E2437 D:\boost\boost_1_32_0\boost/function/function_base.hpp 598(55): 'typename' should be followed by a qualified, dependent type name Does anyone know a workaround for this? Greetings Roman

2 1

RE: [Boost-users] Re: Q: boost::format : can formatting variablea rguments lists be done ?
by Foster, Gareth 07 Mar '05

07 Mar '05

Could you do Logger<DBG> MyLogger; MyLogger.print(); ? -----Original Message----- From: boost-users-bounces(a)lists.boost.org [mailto:boost-users-bounces@lists.boost.org] On Behalf Of Dan Dimerman Sent: 22 December 2004 13:03 To: boost-users(a)lists.boost.org Subject: [Boost-users] Re: Q: boost::format : can formatting variablearguments lists be done ? Richard Howells wrote: > Vladimir is right. > > You've seen the problems and crashes that leaving type checking to the > programmer causes. You need the compiler to do the checking for you. > > I suggest using the same kind of scheme as the iostreams library. You might /.../ > ...it will be (mostly) type safe - the type dangerous stuff is isolated to > one place per output type - and you can extend it as you go. You'll likely > need extra formatting stuff and that's where boost::format may help you. > I was thinking about something like: MyLogger::print( const boost::format& f ) and since all format's operator %() return a boost::format, I could with this single function reap all the types handled by it at once. I use a function rather than the operator << because they are actually template functions, with the parameter being the severity. So for _DEBUG versions, the MyLogger::print< DBG >() function instantiates to a proper call to print, while in _RELEASE versions it instantiates to an empty function. I just didn't see a clean way to do this with operators. Dan _______________________________________________ Boost-users mailing list Boost-users(a)lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

3 2

Program options: terminating parsing.
by The Grumpiest Man You Know 04 Mar '05

04 Mar '05

As context I would like to say that read "The C++ Programming Language" first edition and thought, "That sounds interesting." and then didn't think of it again 'til about a a couple of months ago, when I read Accelerated C++ on the recommendation of a friend and then updated my original copy to the third edition. So I have NO useful C++ experience, please point out even the most obvious things, I really could be that dumb. I have started a non trivial project that I hope will cement some of the things I've read. And that's where boost comes in. I had some success with using boost_filesystem, so I think I am able to extract basic usage out of the documentation but I've failed to find what I'm looking for in program_options. I would really like to parse a line that has the the form: word [options]... nother option-like-non-option-word... so for instance -F after "word" indicates a legitimate value but -F after "nother" shouldn't set or change the value and indeed might not even have a "value" I've been considering producing a class defining operator() to be used as an additional parser that would recognize the second bare word and make all options thereafter into some list for a dummy option, like the file names in the example. However there is mention of allowing multi word options being "not enabled by default" and I thought that the style argument might be the key, but what I've managed to work out from reading a tarball I downloaded of 1_32_0 seems to use style for changing how single words are interpreted. So before I go crazy trying to implement something that sounds easy to say but beyond my ken, is there a method I'm overlooking for terminating the parsing of the command line before the words are all consumed? Thanks for your time, I'll try to be more brief in future. :) -- Blue Skies

3 6

[serialization] problem with Borland C++ Builder 6
by Pablo Aguilar 01 Mar '05

01 Mar '05

Hi, I've just started using the serialization library, and am trying to do so with BC++6. If I'm right, it should work just as long as I remove support for xml archives (because Spirit doesn't work on BC++) I've a class Cliente which has an appropriate serialize member fxn. And a manager class called Clientes which stores a std::vector<boost::shared_ptr<Cliente> > When doing: void Clientes::serialize(Archive& ar, unsigned const version) { ar & m_imp->m_clientes; // where m_imp is a pimpl-style struct // m_clientes is of the type specified above } (this is trimmed down code, I can post the original if required) I get the following error: [C++ Error] serialization.hpp(121): E2015 Ambiguity between 'boost::serialization::void serialize< boost::archive::text_iarchive , boost::shared_ptr<cc::servidor::Cliente> > ( boost::archive::text_iarchive & , _STL::vector<boost::shared_ptr<cc::servidor::Cliente> > & , const unsigned int )' and 'boost::serialization::void serialize< boost::archive::text_iarchive , _STL::vector<boost::shared_ptr<cc::servidor::Cliente> > >( boost::archive::text_iarchive & , _STL::vector<boost::shared_ptr<cc::servidor::Cliente> > & , const unsigned int )' (I removed the allocator parts to make the message clearer) So the question is, how do I remove the ambiguity? Thanks for your help... Pablo Aguilar

3 12