What makes no sense? To be useful, a Parser<T> needs to return a type that is useful for the consumer. If your Parser<T> returns, say, an Id or a unit, and then you combine that with a parser whose type is:
Parser<T> -> Parser<List<T>>
You'll leave your user with the task of filtering their list. If the implementer for the list parser had done the same bad job, you'd use a:
Parser<T> -> Parser<List<T>|()>
And your final type would be:
Parser<List<Id|()>|()>
That would be beyond horrible to use, and anyone using such a parser combinator library would, rightfully, be extremely annoyed.
Parser<T> -> Parser<List<T>>
You'll leave your user with the task of filtering their list. If the implementer for the list parser had done the same bad job, you'd use a:
Parser<T> -> Parser<List<T>|()>
And your final type would be:
Parser<List<Id|()>|()>
That would be beyond horrible to use, and anyone using such a parser combinator library would, rightfully, be extremely annoyed.