Separate compilation is relevant, because it means the version of the interface you compile against may not be the same version you run against. This is fine if the newer version is compatible with the older version. And for the present discussion, we consider an added enum value to not constitute a compatibility break. Nevertheless, it means that the client code can now receive a value that it couldn’t receive before. And it’s useful to be able to define a case distinction for such unknown future values, while at the same time having the compiler check that all currently defined values have been duly considered.

In other words, you want to ensure that you have the most appropriate behavior for whatever values are currently known, and a fallback behavior for the future values that by definition you can’t possibly know at the present time. Of course, this is more or less only practical in languages where the interface version you compile against is only updated deliberately, while the implementation version at runtime can be any newer compatible version.