We need to use the event codes in the event history to determine what events are start, stop or non-calculation events. 

For purposes of this discussion, we are assuming the event history is stored in a coded effective date array called EventHistory, where the start event codes are 10, 11, and 12, and the stop event codes are 31, 41, 51, and 61.

EV_DT:= #START EventHistory &
Code:= #VALUE EventHistory &
; Don't assign the effective dates of the event history into a field called start
; (e.g., Start_DT)! That's confusing because we are working with events that START
; service and events that STOP service; many people would assume that Start_DT
; only contains the effective dates of the start events.

IsStartEvent:= Code #IN (10, 11, 12) & ; i.e., #IN (all the start event codes)
 
IsStopEvent:= Code #IN  (31,41,51,61) & ; i.e., #IN (all the stop event codes)
IsStopEvent:= Code #NOTIN (10, 11, 12) & ; i.e., #NOTIN (all the start event codes)
; If we only have start and stop events (and we don't have to worry about non-service
; related events), the list of the start event codes is smaller than the list
; of stop event codes, so it's simpler to use #NOTIN and list the start event codes

NotAnEvent:= Code #IN (list of non-service related codes) &
NotAnEvent:= #NOT (IsStartEvent #OR IsStopEvent) &
; this is the simplest way if we've already identified the start and stop events
 
Keep in mind, the problem with using codes in an expression is that if someone adds a new code, then someone also has to remember to add the new codes to #IN and #NOTIN in any expression that uses them. Otherwise, the new codes can break these expressions. You could easily add a User-defined "data-based" error message in the Census Specification to check that the event history only contains the expected codes (i.e., someone didn't add a new code).