- Boost-users - lists.boost.org

Re: [Boost-users] Graph Library - bundled properties, internal & external properties
by Elvanör 14 Mar '05

14 Mar '05

Hello again, >> Second, with bundled properties I have troubles with constructors. I >> created a class myClass, that I will use as a struct for the bundled >> properties. This class contains reference members so it needs a >> constructor better than the default constructor without any >> arguments. Creating a graph with these bundled properties doesn't >> allow me to use another constructor than the default myClass() one, >> it seems. I think it would be very good if I could create an array of >> myClass and then point my Graph object to this array to use it as >> bundled properties. Then I could do all the initialization I want >> cleanly. > > I think that when you add an edge or vertex, you can just pass a > myClass instance before the graph argument, e.g,, > > v = add_vertex(myClass(some_reference), g); Ok, so that would solve one of my problems. Thank you. > > >> My last problem is with the depth_first_search() algorithm... and of >> course bundled properties. This algorithm takes a const Graph &; but >> I want to be able to modify (while running the algorithm) the bundled >> properties!! That is the whole point of running this algorithm. I >> don't want to change the structure of the graph (the abstract graph - >> links from vertices to edges, etc), while running the algorithm, of >> course, but I do want to change when I want my bundled properties... >> that is not possible (at least, not without compiler warnings, which >> I don't like :-) due to the fact that depth_first_search expects a >> const graph. > > Yeah, this is an annoying problem. We'd like to be able to take either > a const or a non-const graph, but doing so would require two overloads > of the algorithm, which we really can't do (too much code!). Like someone else pointed out, what do you mean by too much code? It would require too much time? Or would it be considered a problem for more "fundamental" reasons (ie, that could lead to trouble in existing code, etc...)? > I suggest that you put a non-const reference to the graph from within > your visitor and use that instead of the graph that is passed to the > event functions. I am not sure I understand how you would do that... Am I correct in understanding that you would declare a non const reference as a member in your visitor? And then you would have to pass the Graph object when you create the visitor. That means a visitor would be suitable for only one graph (that is not a problem in my case, just a remark) on which you intend to use the visitor on. Another way would be to use pointers to objects as bundled properties, instead of directly the object themselves. I think that should work and that's what I am using right now. Basically it only changes that in your visitor, it's not g[v].some_member = 4; but g[v]->some_member = 4; (if g is the graph). Seems to work fine. Jean-Noël

2 1

Graph Library - bundled properties, internal & external properties
by Elvanör 14 Mar '05

14 Mar '05

Hello everyone, this is my first post to the list so excuse me if I didn't get some concepts right. I have begun to learn the BGL as I was searching for a good graph library; I am a student in Computer Science and my thesis is about graph algorithms. I am impressed, it seems to be a good library. Plus, it has forced me at last to deepen my knowledge of STL, templates and generic programming. Now to my question: as I am a new user, when I read the documentation I decided to use the new bundled properties mechanism rather than the old properties map methods. It was recommended in the documentation (I would, btw, like to humbly point out that the documentation is not very clear on this. The bundled properties explanation page is not accessible from the content page and could confuse quite a lot of people), and anyway I found that a lot simpler. But now I am running into troubles. First, are bundled properties applicable to "external properties"? To be honest, I don't really get the difference between internal and external properties. Or do I have to use property maps (in the old way) for external properties? Second, with bundled properties I have troubles with constructors. I created a class myClass, that I will use as a struct for the bundled properties. This class contains reference members so it needs a constructor better than the default constructor without any arguments. Creating a graph with these bundled properties doesn't allow me to use another constructor than the default myClass() one, it seems. I think it would be very good if I could create an array of myClass and then point my Graph object to this array to use it as bundled properties. Then I could do all the initialization I want cleanly. Right now, it seems to me than the only way to modify these properties (prior to running a grpah algorithm for example), is to do some sort of: G[v].my_first_data_member =5; G[v].my_second_data_member = "test"; etc... with v moving over all the vertexes (or edges). This seems a little bit strange... and impossible to use reference data members with that, for exemple. My last problem is with the depth_first_search() algorithm... and of course bundled properties. This algorithm takes a const Graph &; but I want to be able to modify (while running the algorithm) the bundled properties!! That is the whole point of running this algorithm. I don't want to change the structure of the graph (the abstract graph - links from vertices to edges, etc), while running the algorithm, of course, but I do want to change when I want my bundled properties... that is not possible (at least, not without compiler warnings, which I don't like :-) due to the fact that depth_first_search expects a const graph. I think that will be all... if someone can enlighthen me on these topics I'll be happy!! Regards Jean-Noël

3 3

__stdcall with boost::function
by Olenhouse, Jason 14 Mar '05

14 Mar '05

I'm trying to pass a member function pointer around to the Windows Service Control Manager through Windpws API functions, but I think I'm having troubles with the __stdcall convention. I'm using MSVC 7.1.3088. For brevity I've left out information, but I think this is enough to go on: __ExampleClass.h__ class ExampleClass { protected: void Setup( void ); void __stdcall ServiceMain( DWORD dwArgc, LPSTR *lpszArgv ); LPSERVICE_MAIN_FUNCTION m_fServiceMain; }; __ExampleClass.cpp__ void ExampleClass::Setup( void ) { boost::function<void ( DWORD, LPSTR * )> f2 = boost::bind( boost::mem_fn( &ExampleClass::ServiceMain ), this, _1, _2 ); m_fServiceMain = f2.target<void ( DWORD, LPSTR * )>( ); SERVICE_TABLE_ENTRY dispatchTable[ ] = { { const_cast<char *>( g_sAPP_NAME.c_str( ) ), m_fServiceMain }, { 0, 0 } }; StartServiceCtrlDispatcher( dispatchTable ); } void __stdcall ExampleClass::ServiceMain( DWORD dwArgc, LPSTR *lpszArgv ) { } The error I receive concerns the line m_fServiceMain = f2.target<void ( DWORD, LPSTR * )>( ); error C2440: '=' : cannot convert from 'void (__cdecl *)(DWORD,LPSTR * )' to 'LPSERVICE_MAIN_FUNCTIONA' This conversion requires a reinterpret_cast, a C-style cast or function-style cast LPSERVICE_MAIN_FUNCTIONA is defined in "winsvc.h" and is: typedef VOID (WINAPI *LPSERVICE_MAIN_FUNCTIONA)( DWORD dwNumServicesArgs, LPSTR *lpServiceArgVectors ); Any help or insight would be much appreciated. -Jason

3 2

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

[wiki] New content filtering to reduce spam
by Jeff Garland 14 Mar '05

14 Mar '05

All - I've instituted a new content filter on the Wiki to try and cut off the spamming without going to a login or other user intrusive mechanism. The timing of this feature turned out to be important as a sophisticated spambot that evades the ip banning strategy by using many different ip addresses (presumably using zombie machines) has been attempting to SPAM the wiki all day today. The new content filter has sucessfully stopped this bot. Note that if a spammer is successful I am now reverting all spam content from the Wiki history so that there is no chance of googlebot finding it. For the moment, I'm not going to publish the regular expressions I'm using to filter, but suffice to say, normal boost content should be unaffected. If you get an error indicating content has been blocked due to spam, send me and email. Thanks, Jeff

1 0

How does Boost.Build/bjam find boost?
by Kelly Burkhart 13 Mar '05

13 Mar '05

Hello, I am starting to fiddle with boost and bjam and am not able to get bjam to find the boost headers and libraries. I'm running SuSE 9.1 and started out using the RPMs. Since last night, one-by-one I've uninstalled the RPMs and replaced with freshly downloaded and build components. Right now I have the latest boost,bjam and boost.build releases. I built boost with --prefix=/opt/boost and installed it. I have boost.build placed in /opt/boost as well. Under my home directory I have a directory in which I'm trying to build a hello-world program. The directory contents: > ls boost-build.jam Jamfile project-root.jam tst.cpp > cat boost-build.jam boost-build /opt/boost/boost-build ; > cat Jamfile exe tst : tst.cpp ; > cat tst.cpp #include <iostream> #include <string> #include "boost/format.hpp" int main() { std::cout << boost::format("%1%, world") % "hello" << std::endl; return 0; } project-root.jam is empty. When I try to build with 'bjam tst' it is not able to find the boost headers, which makes sense, because I've not been able to figure out how to indicate where they live. How is this accomplished? -K

3 9

Boost::regex: Removing Forte CC compiler warnings
by Rajesh Radhakrishnan 13 Mar '05

13 Mar '05

Hi, Any suggestions on removing these compiler warnings (CC: Sun C++ 5.5 2003/03/12 ) : /usr/local/boost_regex/include/boost/type_traits/is_convertible.hpp", line 279: Warning (Anachronism): Type names qualified by template parameters require "typename". "/usr/local/boost_regex/include/boost/type_traits/is_convertible.hpp", line 279: Where: While specializing "boost::is_convertible<boost::From, boost::To>". "/usr/local/boost_regex/include/boost/type_traits/is_convertible.hpp", line 279: Where: Specialized in non-template code. "/usr/local/boost_regex/include/boost/type_traits/is_convertible.hpp", line 279: Warning (Anachronism): Type names qualified by template parameters require "typename". "/usr/local/boost_regex/include/boost/type_traits/is_convertible.hpp", line 279: Where: While specializing "boost::is_convertible<boost::From, boost::To>". "/usr/local/boost_regex/include/boost/type_traits/is_convertible.hpp", line 279: Where: Specialized in non-template code. Line 279: /usr/local/boost_regex/include/boost/type_traits/is_convertible.hpp BOOST_TT_AUX_BOOL_TRAIT_DEF2(is_convertible,From,To,(::boost::detail::is_convertible_impl<From,To>::value)) -- Regards Rajesh

2 1

MPL - creating new sequence from another sequence using predicate
by Bruce Trask 13 Mar '05

13 Mar '05

Hi, I have a mpl::vector of types. Each of the types in the vector have themselves another nested mpl::vector that is a sequence of enums. I have include the code below. What I am trying to do is take the original sequence and make a new sequence based on the predicate that only elements with certains enums in their nested vector be included in the new sequence. I have included more notes in the embedded comments. I was thinking of using mpl::copy_if but am having trouble getting the predicate right. Wanted to know if anyone could steer me in the right direction. I am trying to fill in the ??? below. I am having trouble writing a custom predicate. Thanks in advance. Regards, Bruce #include <string> #include <iostream> #include <boost/mpl/vector.hpp> #include <boost/mpl/begin_end.hpp> #include <boost/mpl/deref.hpp> #include <boost/mpl/count_if.hpp> #include <boost/mpl/size.hpp> #include <boost/mpl/for_each.hpp> #include <boost/type_traits/is_member_pointer.hpp> #include <boost/mpl/vector_c.hpp> #include <boost/mpl/copy_if.hpp> #include <boost/mpl/less.hpp> #include <boost/mpl/placeholders.hpp> #include <boost/mpl/find.hpp> namespace mpl = boost::mpl; using namespace mpl::placeholders; enum colors { red, blue, green, yellow }; template<typename COLORTYPES> struct TestClass { typedef COLORTYPES colortypes; }; void func() { // come up with a interface to simplify this like vector_c - only it must be able to handle enums - I would think one would already exist typedef mpl::vector< mpl::integral_c<colors,red> , mpl::integral_c<colors,green> , mpl::integral_c<colors,blue> > colorcoll1; typedef mpl::vector< mpl::integral_c<colors,red> , mpl::integral_c<colors,blue> > colorcoll2; typedef mpl::vector< mpl::integral_c<colors,green> > colorcoll3; typedef mpl::vector<TestClass<colorcoll1>, TestClass<colorcoll2>, TestClass<colorcoll3> > testvector; std::cout << " SIZE IS " << mpl::size<testvector>::value << std::endl; // would like to make a new vector from testvector that contains only those types that have green in their nested vector colortypes. //In other words, go from testvector which has three types to a vector that only has TestClass<colorcoll1> and TestClass<colorcoll3> // typedef mpl::copy_if<testvector, ???? >::type testvectorwithgreen; // I can do it successfully when operating directly on the container as below, but am having trouble deterimining the predicate for the above case typedef mpl::vector_c<int, 1,2,3,4,5,6,7,8> intColl; std::cout << mpl::size<intColl>::value << std::endl; typedef mpl::copy_if<intColl, mpl::less<mpl::_1, mpl::int_<5> > >::type newIntColl; std::cout << mpl::size<newIntColl>::value << std::endl; } int main() { func(); return 0; }

3 4

boost::fsm, parameters to DestinationState
by Jan Eickmann 12 Mar '05

12 Mar '05

Hi, when doing a transition using transit<myDestinationState>, I would like to pass some parameters to myDestination. Problem is, those parameter-values will not be known at compile time, so I can't just pass them in as template-parameters to myDestinationState. Of course I could just set some global (or static member member) variable and read that from myDestinationState's constructor but this doesn't feel like a very clean design. Any suggestions? Jan Eickmann

2 6

base64->wstring decoding
by Stefan Strasser 12 Mar '05

12 Mar '05

Hi, I need to write a std::wstring which can contain binary data like \0 to a xml attribute, so I'm converting it to base64 but I have problems decoding it: typedef transform_width<binary_from_base64<std::string::const_iterator>,32,6> decode; std::wostringstream out; std::copy(decode(in.begin()),decode(in.end()),std::ostream_iterator<wchar_t,wchar_t>(out)); is this a correct way to decode base64? it works, but only if (bits % 6 == 0). otherwise the program is abort()ed due to an exception. the string was encoded by base64_from_binary<transform_width<std::wstring::const_iterator,6,32> > Thanks, -- Stefan Strasser

2 2