Sorry. This page wasn't really written as an introduction; more a collection of ideas. To break that example down for you, the outer {} indicate that this is a

Sorry. This page wasn't really written as an introduction; more a collection of ideas. To break that example down for you, the outer {} indicate that this is an anonymous function. It is implicitly a function over a single argument named "x"; our list of dictionaries. Expressions read right to left, so the first semantic unit is: k:?,/!:'x The variable "k" gets the distincts (?) of the concatenate-reduction (,/) of the keys (!:) of each (') dictionary: !:'x (,`a ,`b `a `b) ,/!:'x `a `b `a `b ?,/!:'x `a `b The other embedded lambda "{x@&~_n~/:x}" is read as "x indexed (@) by where (&) not (~) null (_n) matches (~) each-right (x)"; it filters nulls from a list. _n~/:(22;;33;55) 0 1 0 0 ~_n~/:(22;;33;55) 1 0 1 1 &~_n~/:(22;;33;55) 0 2 3 {x@&~_n~/:x}(22;;33;55) 22 33 55 Then the overall construction "{x@&~_n~/:x}x@'/:k" can be understood as "reach into each dictionary with all the keys and filter them to non-null values". Now you have a list of lists of values in the order that the keys appeared. Finally, ".+(k; ... )" joins our keyset with the values we found for each key. Make dictionary (.) from the transpose (+) of the pair of vectors (k;v). Does that help?