Peter Achten
In my opinion a programmer should not have to register/unregister callbacks explicitly.
I don't understand what you're proposing in its place. You say that on X you do need to [un]register callbacks and on Windows you use a filter function to determine whether or not to use the callback. Either way, it seems that there is some mechanism which lets the backend (windows, X, Motif, etc) know that you no longer need to receive events of that type so the API is going to have to provide appropriate functions. That is, there will be a way to register to receive those events and a way to stop receiving those events. Whether they register and unregister a callback or change a bitset defining which events to receive or do something else is largely irrelevant. -- Alastair Reid alastair@reid-consulting-uk.ltd.uk Reid Consulting (UK) Limited http://www.reid-consulting-uk.ltd.uk/alastair/ ps The discussion of the cost of receiving mouse events seems to provide an argument against using streams of events as the base on which other APIs are implemented. If programs explicitly register for events/callbacks, then we can avoid the cost of propagating events that aren't wanted. If the GUI simply spouts a stream of timestamped event objects as in Fran-like systems, then we have to propagate events which aren't needed because there's no easy way to know what events are wanted. (A stream-based system could have operations to turn certain events on and off but this leads to a difficult programming style where you have to coordinate your pure event-stream amnipulation code with your monadic event on/off code.) This isn't to say that streams are a bad model simply that they can't express something that the monadic models can.