The EBNF grammar definition in the kata is a little off. query includes join as [ ws, join ], but it's also included in the definition of from: [ { ws, join } ]. Seeing as the examples include a query with multiple joins, I believe the inclusion in query's definition is in error and should be removed.

For those attempting this Kata without outside help, assuming all points are part of the hull and then reducing them is the WRONG approach. It will work, but it's not performant enough (at least for some languages) and you will time out.

Can we get some leniency with precision on the random tests (haskell)? I'm coming up with the correct answer, but the last one or two digits are always off from the expected value.