Hi, with the unit test added by Stefan it is pretty good reproducable with VS2008/2010 and 32bit at least on my machine. The compiler actually throws a warning that there might be something fishy:
http://msdn.microsoft.com/en-us/library/1s6193tt(v=vs.80).aspx As I did not see this warning ever in my life, I just started investigating what's the deal with that. Basically, it's a problem of member function pointers used within (multiple) inheritances. As there is no proper support for member functions pointers being a real C++ feature like a delegate in other languages, it's up to the compiler to deal with that. It even gets more complicated if virtual functions or virtual inheritance is involved... For further reading: http://social.msdn.microsoft.com/Forums/en-US/vclanguage/thread/05a59a5f-1370-418f-9460-e4dd49d19ddf/ http://www.codeproject.com/Articles/7150/Member-Function-Pointers-and-the-Fastest-Possible So it's up to the compiler vendors to handle this issue appropriately. Of course, Microsoft has some "solutions" for this: http://msdn.microsoft.com/en-us/library/ck561bfk%28VS.80%29.aspx Imo, this not really practical, so the only portable solution is reordering the interheritance, meaning the base class used for the function binding (aka dispatcher in this case) has to be very first. A remaining question on this is what gcc for instance is doing about this whole issue? A solution on the library side would be thinking about using something similar like boost.function, which from what I read so far should have no problem with this particular problem, as boost.bind used for creating the actual binding is due its implementation aware of the complete class hierarchy. Cheers, Daniel _______________________________________________ eq-dev mailing list [email protected] http://www.equalizergraphics.com/cgi-bin/mailman/listinfo/eq-dev http://www.equalizergraphics.com
