If the instance-specific behaviour differs in just one or two places, and there are only a couple of enumeration values, I'll just use ifTrue:/ifFalse:
method1
self myEnumValue == #firstEnum ifTrue: [self doTheFirstKind] ifFalse: [self doTheSecondKind].
method2
self myEnumValue == #firstEnum ifTrue: [self againItsTheFirstKind] ifFalse: [self doTheSecondKind].
If there are a few more enum values I'll use a dictionary, and/or start forming method names from the enum value and using perform:
method1
self perform: self myEnumValue
and if that starts to feel ugly, or there are more places that depend on the enum value, I'll use an abstract class and a subclass for each enum value. It wouldn't necessarily be a singleton subclass: if the enum is only used as an instance variable of one existing class, that class can be made abstract, the instance variable removed, and a subclass made for each enum value.
Personally I find that adding a set of classes always feels rather heavyweight. I know they're only objects, but the development tools give them a strong visibility, with no good way to highlight "real" classes over "Enum1, Enum2, Enum3" mini-classes.
Steve
Locked
