>Then the design is simply wrong. No program, of any kind, should destroy data without warning and with no reasonable expection that this would happen. I know of no other program where data is destroyed simply by renaming an object.

Mark,

I disagree with your conclusion. This design is fine with me. Only the class to which the method was added knows about the existance of that method. If you change the name of that method in the class to which it was added, you take the responisiblity for all of the sub-classes and their dependence on that method.

>Right, I understand this. What I expected to happen was the the child class method would remain in the child class under the original name. The fact that this doesn't happen is a design bug.

The child class NEVER had that method in it. The method was in the parent class and you changed it there, the child simply inherited from its parent. When the parent no longer has a method named Foo, then the child class no longer has a method named Foo also.