Exponential Backoff Idiom

[Eric Newhuis]

Has anyone a better version of this they'd like to share?  The trouble with this approach is that Fun cannot block.  Maybe there is an entirely better way to express this so it works as a mix-in concept whether or not Fun has its own receive loop.  I wrote this expecting it would not.

backoff(N, Multiplier, Max, Fun, Acc) ->
		case Fun(Acc) of
				{continue, Acc2} ->
						receive
						after
								N ->
										discover(?IF(N < Max, N*Multiplier, Max))
						end;
				{ok, Acc2} ->
						Acc2
		end.