Why namespaces are not as good as classes

grid.clj

(ns mark.grid)

; I'm working with grids that are primarily 3x4.  
; I get up and running with the following simple code.

(def rows 3)
(def cols 4)

(def cells (for [row (range rows), col (range cols)] [row col]))

(defn select-cells-by-row [row]
  (filter (fn [[r c]] (= row r)) cells))

(defn print-cells-by-row [row]
  (doseq [cell (select-cells-by-row row)]
    (println cell)))

; But now, I realize that I want to work with other size grids.
; If this namespace were a "class", the change would be trivial:
; just add a constructor that sets the rows and cols

; In Clojure, we end up with this kind of nonsense:

(defrecord GridSize [rows cols])
(def default-grid-size (->GridSize 3 4))

(defn cells
  ([] (cells default-grid-size))
  ([{rows :rows cols :cols}]
    (for [row (range rows), col (range cols)] [row col])))

(defn select-cells-by-row
  ([row] (select-cells-by-row default-grid-size row))
  ([grid-size row]
    (filter (fn [[r c]] (= row r)) (cells grid-size))))

(defn print-cells-by-row
  ([row] (print-cells-by-row default-grid-size row))
  ([grid-size row]
    (doseq [cell (select-cells-by-row grid-size row)]
      (println cell))))

; IMHO, this is an obscene amount of additional code to scale
; a project that relies on mutual functions seeing the same
; vars to one in which those vars become parameters.

Hi Mark, maybe you want this sort of thing? https://gist.github.com/johnwalker/8142143

(The idea being that you can store the methods according to a protocol referencing the records parameters.)