Parsers are working in a unique buffered area (the file to parse is
entirely loaded into this unique buffer during parsing). This solution make
it easy point/line/column counting as well as backward moves.
The acces to the current character (feature cc) is also very fast.