Combinatory parsing fascinates me. It is an area that has open problems, but also a lot of solved problems that make great exercises. A while ago I tried to define a combinatory regular expression matcher in F#, and was puzzled at how complex it turned out to be. As usual, marginal understanding was to blame.
Now I turned to Haskell for a change. I do think it helped, to learn more both about the problem and about Haskell. I hope to backport this exercise to the ML world, in the meanwhile here is the gist: