Return to Revisions

To be frank: Your implementation does not target the topic token scanning.
It is targeted at serialization / deserialization.
But the example that it is targeted at is no example of token scanning neither.
Tokens in the example would be the '.', [0-9]+, token scanning would first
build a token tree made of this tokens before interpreting the tokens in a
grammatical context. The implementation you choose is very clear. For the example used it could be irritating to the reader as it introduces other non - trivial concepts but I personally tend to teach clear and structured programming over examples that lead to bad code design (and mark it as "is extremely crude" ;)
I personally would not recommend recursion in operator implementations though, specially not in stream operators. You would not want to call a stream operator and get an error such as "too many recursions".