Ha, you gonna love this :)
Its all about performances;)
Lagarto parser emits all the text as CharSequence. By default, we are using CharBuffer for implementation. If the CarSequence interface works for you, then you are set - and you will have great performances.
However, sometimes CharSequence is not enough. For example, to build our DOM we store all the texts as Strings. Now, we could do `toString` on the char buffers, but overall that is less performant - because we are first creating char buffer and then strings. Therefore, if you emit strings (set it to true), the implementation for `CharSequence` is going to be a String, and calling toString() on it (to get String type) will have no performance penalties.
That flag determines the implementation of CharSequence, thats al :)
And you are right, this is important, should go to the docs ;)