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

[BGL] connected components algorithm
by Alejandro Aragon 05 Aug '06

05 Aug '06

Hello everyone. I'm trying to implement Fleury's algorithm to find an Eulerian path in a graph. Every time I erase an edge in the graph, I need to check how many connected components I have (because I can't erase an edge that is a "bridge"). Also, I was using in the beginning graphs with VertexList=VecS but it turns out that when you erase an edge all the vertex descriptors are invalidated. Thus, I had to copy my graph to another one with VertexList=listS. I was wondering if the connected components algorithm works with this type of graph. I'm having an error. It seems that the compiler does not recognize this function. /usr/include/boost/graph/connected_components.hpp:46: error: no matching function for call to ‘put(unsigned int*&, void*&, unsigned int&)’ make: *** [graph.o] Error 1 Any ideas? Thank you, aa My code is shown below: typedef adjacency_list<listS, listS, undirectedS, property<vertex_index_t, std::size_t>, no_property > graph_t; graph_traits<graph_t>::out_edge_iterator out,out_end; graph_t copy(num_vertices(*gPtr)); // obtain a property map for the vertex_index property property_map<graph_t, vertex_index_t>::type index = get(vertex_index, copy); // initialize the vertex_index property values graph_traits<graph_t>::vertex_iterator vi, vend; graph_traits<graph_t>::vertices_size_type cnt = 0; for(tie(vi,vend) = vertices(copy); vi != vend; ++vi) put(index, *vi, cnt++); for (tie(ei, ei_end) = edges(*gPtr); ei != ei_end; ++ei){ add_edge(vertex(source(*ei,*gPtr), copy), vertex(target(*ei,*gPtr), copy), copy); } graph_traits<graph_t>::vertex_descriptor Vertex; graph_traits<graph_t>::edge_iterator ei,ei_end; vertex_descriptor u,v; u = vertex(0,copy); for (tie(out, out_end) = out_edges(u,copy); out != out_end; ++out){ u = source(*out,copy); v = target(*out,copy); remove_edge(u,v,copy); std::vector<size_t> component(num_vertices(copy)); num = connected_components(copy, &component[0]);

2 2

Jim Lear is out of the office.
by Jim.Lear＠legerity.com 05 Aug '06

05 Aug '06

I will be out of the office starting 08/05/2006 and will not return until 08/12/2006. I will respond to your message when I return.

1 0

assert when using cregex_iterator
by Winson Yung 05 Aug '06

05 Aug '06

I have a very simple test program using the latest 1.33 version of regex library. As soon as I run it, it assert (crash) at the line where std::string s = (*m1)[0]. I am doing something wrong? ============== #include "stdafx.h" #include <boost/regex.hpp> const char matchpattern[] = "\\s+kitty this"; const char astring[] = "hello Kitty this is a test"; using namespace boost; int main(int argc, char* argv[]) { regex header_expr(matchpattern, regex_constants::basic | regex_constants::icase); cregex_iterator m1(astring, astring + strlen(astring), header_expr); std::string s = (*m1)[0]; <==== crash here return 0; }

2 1

boost python noobie questions
by chun ping wang 05 Aug '06

05 Aug '06

Hey I have several problem... A.) Can boost::python work with templated argument... Example: /************************************************ * rsa(p, q): * * let n = p * q * * let totient = (p-1)*(q-1) * * choose integer e between 1 and totient * * such that greatest common denominator is 1 * * d = (1 mod totient) / e * * return e, d, n * ************************************************/ template <class T> boost::tuple<T, T, T> boost::rsa_generate(const T& p, const T& q) { BOOST_STATIC_ASSERT(boost::is_integral<T>::value); assert(p != q); const T n(p*q); const T totient((p-1)*(q-1)); T e(3); for (e; EuclidBinaryGCD(e, totient) != 1; e = Random::randint(T(2), totient-1)); const T d(Inversemod(e, totient)); return boost::make_tuple(e, d, n); } BOOST_PYTHON_MODULE(rsa) { def("rsa_generate", boost::rsa_generate); } Second question.. how come lambda function don't work. COmpling this by itself..give no error. template <class T, template <typename ELEM, typename = std::allocator<ELEM> > class CONT> double MyAnn::getCondProb(const T& A, const T& B, const CONT<typename std::pair<T, T> >& DATA) { using namespace boost::lambda; typedef typename std::pair<T, T> MyPair; // gets the number of B const int numOfB ( std::count_if(DATA.begin(), DATA.end(), bind(&MyPair::second, _1) == B ) ); // gets the number of A given B. const int numOfAGivenB( std::count_if(DATA.begin(), DATA.end(), (bind(&MyPair::second, _1) == B) && (bind(&MyPair::first, _1) == A) ) ); return (numOfAGivenB*1.0)/numOfB; } now including this file.. and compling this gives error about "_1" BOOST_PYTHON_MODULE(bayesian) { def ("getCondProb", getCondProb<long long>); } Thanks for the help. If its the way i compile.. how i would i get it to work.. iam using dev-cpp 4.9

2 2

Boost.Signals example/doc_view application question
by Allen Gooch 04 Aug '06

04 Aug '06

I have a quick question regarding connecting a bound member function. In the doc_view example application signals are defined as requiring a function taking a bool and returning nothing. class Document { public: typedef boost::signal< void (bool) > signal_t; // ... void append(const char* s) { m_text += s; m_sig(true); } // ... }; The View abstract base class requires concrete bases to provide a concrete implementation of the refresh() member function matching the slot signature required. class View { public: virtual void refresh(bool) const = 0; // ... }; Here's my problem. If I change the signal to take no args using the preferred or portable syntax I get compilation errors: class Document { public: typedef boost::signal< void () > signal_t; // ... void append(const char* s) { m_text += s; m_sig(true); } // ... }; class View { public: virtual void refresh() const = 0; // ... }; Compilation output (msvc-7.1): Compiling... SignalTest.cpp c:\Boost\include\boost-1_33_1\boost\bind.hpp(63) : error C2825: 'F::result_type': cannot form a qualified name c:\Boost\include\boost-1_33_1\boost\bind\bind_template.hpp(15) : see reference to class template instantiation 'boost::_bi::result_traits<R,F>' being compiled with [ R=boost::_bi::unspecified, F=void (__thiscall View::* )(void) const ] c:\development\SignalTest\SignalTest.cpp(49) : see reference to class template instantiation 'boost::_bi::bind_t<R,F,L>' being compiled with [ R=boost::_bi::unspecified, F=void (__thiscall View::* )(void) const, L=boost::_bi::list2<boost::_bi::list_av_2<View *,boost::arg<1>>::B1,boost::_bi::list_av_2<View *,boost::arg<1>>::B2> ] c:\Boost\include\boost-1_33_1\boost\bind.hpp(63) : error C2039: 'result_type' : is not a member of 'operator``global namespace''' c:\Boost\include\boost-1_33_1\boost\bind.hpp(63) : error C2146: syntax error : missing ';' before identifier 'type' c:\Boost\include\boost-1_33_1\boost\bind.hpp(63) : error C2955: 'boost::_bi::type' : use of class template requires template argument list c:\Boost\include\boost-1_33_1\boost\bind.hpp(112) : see declaration of 'boost::_bi::type' c:\Boost\include\boost-1_33_1\boost\bind.hpp(63) : fatal error C1903: unable to recover from previous error(s); stopping compilation Observed with boost_1_33_1 release on cygwin/gcc-3.4.4 and msvc-7.1 platforms. Any ideas? Many thanks in advance, -allen

1 0

Re: [Boost-users] Ownership of Boost IP?
by Nat Goodspeed 04 Aug '06

04 Aug '06

> -----Original Message----- > From: boost-users-bounces(a)lists.boost.org [mailto:boost-users- > bounces(a)lists.boost.org] On Behalf Of Paul Giaccone > Sent: Friday, August 04, 2006 7:38 AM > To: boost-users(a)lists.boost.org > Subject: Re: [Boost-users] Ownership of Boost IP? > > Nat Goodspeed wrote: > > I suppose the next round would involve my arguing that it would take me > > many months to implement even part of the functionality in question > > > The answer to that would be to contact the Boost developers for all the > libraries you intend to or might use, and ask them how long it took them > to code up their respective libraries. [Nat] I'm hoping to avoid any approach that would add per-Boost-library overhead. Having to decide in advance the set of libraries I intend to use - and later facing additional internal battles to get permission to use one more -- would be discouraging. But I take your point: you're talking about an example set, not literally suggesting that I repeat the process every time I desire to incorporate another new Boost library. > Add it all up and present that > figure to your boss, pointing out that these people were at least as > competent as you or any other programmer he might care to employ. Does > he *really* want to wait a year or two for the code you are writing for > him? [Nat] Certainly he wouldn't want that. I guess I wasn't thinking clearly last night. The real problem is that nobody -- not me, not my boss, not the lawyers -- wants me to (attempt to) reimplement any Boost library. If we can't use Boost, then we must resort to the ugly hand-coded mechanisms that C++ programmers used before Boost emerged. No smart pointers? Delete the !@#? pointers yourself! No bind()? No boost::function? Quit griping and use the STL adapters! Can I make a case that the small incremental costs in coding time and reliability would add up to a bigger hit than the hypothetical risk of lawsuit? It seems doubtful, since all the factors involved in such an argument are sheer guesswork -- and every party will believe their own guesses rather than their opponents'. My boss might even be sympathetic. But my boss isn't the real problem here. > You might also want to ask your boss whether he is concerned that > Microsoft (or whoever wrote your compiler) might sue for use of their > version of C++, or whether he thinks Bjarne Stroustrup might be sending > you a writ... Of course, that would just be ridiculous, wouldn't it? ;) [Nat] Microsoft it is. I think I heard that Microsoft's licenses (purport to) protect you, the user of their products, from this kind of liability. That may be what raised this whole issue in our legal dept. in the first place.

2 1

Re: [Boost-users] Ownership of Boost IP?
by james.jones＠firstinvestors.com 04 Aug '06

04 Aug '06

From: "Nat Goodspeed" <ngoodspeed(a)solidworks.com> > As the lawyers' position is "Don't use > Boost," I can't see them making that effort, so it would be up to me (or > other interested developers) to pursue it. Which leaves me/us vulnerable > to charges of "wasting time" that could "better" be spent reinventing > Boost technologies. How about mentioning that a bunch of Boost stuff (ideas, at least) is intended for use in TR1 and, AFAIK, TR2 and TRn. And that surely standard libraries wouldn't be subject to these IP concerns? (Of course, they might reply: "Fine, once it gets into standard libraries you can use it." Which means you're waiting years.) - James Jones Administrative Data Mgmt. Webmaster 375 Raritan Center Pkwy, Suite A Data Architect Edison, NJ 08837

4 3

Re: [Boost-users] Ownership of Boost IP?
by Sohail Somani 04 Aug '06

04 Aug '06

> -----Original Message----- > From: boost-users-bounces(a)lists.boost.org > [mailto:boost-users-bounces@lists.boost.org] On Behalf Of Robert Ramey > Sent: Friday, August 04, 2006 9:52 AM > To: boost-users(a)lists.boost.org > Subject: Re: [Boost-users] Ownership of Boost IP? > > > Realistically, this situation isn't about the facts - its about > someone having authority beyond his expertise. Its a > political issue not a technical one. Boost and its developers > can only do so much. Therefore, it can all be solved with a nice lunch with the boss! Bosses are good at this stuff. Atleast the good ones :)

1 0

Re: [Boost-users] Ownership of Boost IP?
by Nat Goodspeed 04 Aug '06

04 Aug '06

> -----Original Message----- > From: boost-users-bounces(a)lists.boost.org [mailto:boost-users- > bounces(a)lists.boost.org] On Behalf Of Robert Ramey > Sent: Friday, August 04, 2006 11:18 AM > To: boost-users(a)lists.boost.org > Subject: Re: [Boost-users] Ownership of Boost IP? > > Nat Goodspeed wrote: > > My employer's legal department has just stated that we may not use > > Boost for product development because of a hypothetical lawsuit risk. > > Hmmm - wouldn't his same concern apply to any software you > don't write yourself? Suppose you use VC 7.1. It comes with > dinkumware libraries - does your legal dept require a release > from dinkumware?. Assuming it does, how does assure itself > that some dinkumware developer didn't use some code/knowledge > from his previious job or was inspired by some snippet he found > on the net. OK so you get releases and hold harmless clauses > from all the past dinkumware developers. Now how about the > windows API - after all that's just another library. Are you going > to use calls from that? Does your legal department feel confident > that no microsoft employee or manager has incorporated code > from other sources? If it does, its pretty clear they've not aware > of all the lawsuits microsoft has had in this regard. And the fact > is that none of the lawsuits have resulted in damages to microsoft > customers or users or developers' who have leveraged on such > code. [Nat] To play devil's advocate for a moment -- I'm sure the legal dept. would say that's because Microsoft promises protection from such actions, in return for the license fee. If I could show precedent where someone attempted to sue users of (or developers who integrated) an Open Source project -- but lost -- that would be a stronger argument. > Such a decision has technical and economic implications far, far > beyond what your legal department (and probably management - if its gotten > this far) are aware of. [Nat] No argument from me there. > Maybe more practical advice would be > to suggest to your boss that he get a second legal opinion. [Nat] Interesting suggestion! I'll see if I can make anything happen along those lines.

2 1

Re: [Boost-users] shared_ptr in API
by Sohail Somani 04 Aug '06

04 Aug '06

If the pointer is meant to be owned by the caller, you could consider references (assuming the objects are required of course). > -----Original Message----- > From: boost-users-bounces(a)lists.boost.org > [mailto:boost-users-bounces@lists.boost.org] On Behalf Of > Mikhail Kubzin > Sent: Thursday, August 03, 2006 3:45 AM > To: boost-users(a)lists.boost.org > Subject: Re: [Boost-users] shared_ptr in API > > Did you consider using weak_ptr in your API? > > > > _______________________________________________ > Boost-users mailing list > Boost-users(a)lists.boost.org > http://lists.boost.org/mailman/listinfo.cgi/boost-users >

1 0