I've assembled a few parsers using various tools, and I must say PEGs are my favorite. They are very flexible and performance is comparable. (even CoffeeScript 2.0 is being written with a PEG.)
Ah, ok cool. I haven't tried using PEGs and I don't even know the theory behind it, so I'm watching a video [1] by the author of LPEG now, which is quite good. Cheers!
[1] http://vimeo.com/1485123