“Interface” like semantics with boost::bind?
Boost::bind takes its parameters by value and copies them. That means f = boost::bind(boost::mem_fn(&X::go), x) will pass a copy of x which will slice off the Y piece of it (if it was really a Y to begin with). To get virtual dispatch to work, you need to pass a pointer to boost::bind : f = boost::bind(&X::go, &x) (Note that you don't actually need mem_fn or to explicitly write void since boost::bind and argument deduction take care of those for you.).
Boost::bind takes its parameters by value and copies them. That means f = boost::bind(boost::mem_fn(&X::go), x); will pass a copy of x, which will slice off the Y piece of it (if it was really a Y to begin with). To get virtual dispatch to work, you need to pass a pointer to boost::bind: f = boost::bind(&X::go, &x); (Note that you don't actually need mem_fn, or to explicitly write , since boost::bind and argument deduction take care of those for you.).
Java interfaces don't specifically exist within C++. Closest you can get is pure abstract base classes. This is generally quite close enough.
The rest of your question is unrelated to interfaces. Java uses the Observer pattern for event connection and dispatch. The interface part is only mildly related because observers are required to obey specific interfaces (of course, since otherwise you wouldn't have any idea what to call).
Using boost::bind to create functors is actually an abstraction beyond interfaces and is thus a more generic solution. The observer pattern and functors are put together into signal/slot idiom/patterns implemented in various libraries like boost::signals, boost::signals2, and gtk++. The Qt version is quite different in mechanics but similar in concept.So, what's this mean to help you understand what, why and where?
I'd suggest starting with a search on what the Observer pattern is and try to write a few implementations.
I wanted to be able to have something like Java's interface semantics with C++. At first, I had used boost::signal to callback explicitly registered member functions for a given event. This worked really well.
But then I decided that some pools of function callbacks were related and it made sense to abstract them and register for all of an instance's related callbacks at once. But what I learned was that the specific nature of boost::bind and/or taking the value of this seemed to make that break. Or perhaps it was just the fact that the add_listener(X &x) method declaration changed the code that boost::bind generated.
I have a very rough understanding why the problem occurred and I think it's probably functioning correctly as per its design. I am curious: what should I have done instead?
I cant really gove you an answer,but what I can give you is a way to a solution, that is you have to find the anglde that you relate to or peaks your interest. A good paper is one that people get drawn into because it reaches them ln some way.As for me WW11 to me, I think of the holocaust and the effect it had on the survivors, their families and those who stood by and did nothing until it was too late.
Related Questions
How to alter a queue table's column from byte semantics to character semantics (Oracle)?
Problem with boost::bind, boost::function and boost::factory?
Copy constructor not called when using boost::bind, boost::shared_ptr, and boost::thread together?
Boost bind inside Boost packaged_task. Why boost asio thinks its not CompletionHandler?
How can I use boost::bind to bind a class member function?
Boost::lambda bind expressions can't get bind to string's empty() to work?