Is this a completely new problem? f(x+ y) is definitely NOT equal to f(x)+ f(y) for the problem you gave before. For example, [itex]1+ \sqrt{2}[/itex] is irrational and so [itex]f(1+ \sqrt{2})= 1- (1+ \sqrt{2})= -\sqrt{2}[/itex] but since 1 is rational and [itex]\sqrt{2}[/itex] is irrational, [itex]f(1)+ f(\sqrt{2})= 1+ (1- \sqrt{2})= 2- \sqrt{2}[/itex].
You didn't say anything about using [itex]\epsilon[/itex] and [itex]\delta[/itex] in your first post. If you are not allowed to use "[itex]\lim_{x\to a} f(x)= L[/itex] if and only if, for any sequence [itex]\{x_n\}[/itex] that converges to a, the sequence [itex]\{f(x_n)\}[/itex] converges to L", then copy the proof of that theorem, for this particular function.