Currently, if users are iterating a ChangeStream, they have to check both stop conditions Err() != nil (i.e. "there was an error") and if ID() == 0 (i.e. "the server invalidated the cursor"). For example:

for {
	if c.TryNext(...) {
		// Use the result.
	}
	if c.Err() != nil {
		panic(c.Err())
	}
	if c.ID() == 0 {
		break
	}
}

A more obvious pattern is to return an error if the server invalidates the cursor. In that case, users only need to check Err() != nil.

Open questions:

• Are users still able to differentiate between errors and server-invalidated change stream cursors (i.e. will ID() only return 0 for invalidated change streams, or any error)?
• Should we implement the same behavior for tailable cursors?