On Tue, Oct 04, 2005 at 01:41:42PM -0500, Ben Loftis wrote: > > Interesting. I'll have to go through some mental exercise to convince > > myself that this will never cause infinite cycles. > > The controllers should not send a change back to the M when they are told by > the M what their value is.
The whole point about Paul's scheme is that he *is* doing exactly that. And while does seem odd, I will not dismiss the idea without giving it a full-length meditation :-) > You'll also want some logic in the M that says "if I receive a change that > tells me to go to my current value, then don't send updates to anyone". This should be a rare occurence, so I wonder if it's worth the extra code. What could be important when the client is a GUI is to limit the number of updates. Dragging a slider could result in *a lot of* X move events, and it's probably unwise to try and update the model and all clients for each one. -- FA
