Proposal -distribute --- Distribute a set of objects among different bins

[emacksnotes]

Proposal -distribute --- Distribute a set of objects among different bins; each bin is associated with a predicate function. In other words, an object belongs to a bin, if it satisfies the corresponding bin function.

See also

• collect-cond! a variant of -separate and
• -separate-multi

(defun distribute (data bin-functions)
  (let* ((f (lambda (x)
              (->> bin-functions
                   (--some (when (funcall it x)
                             it))))))
    (->> data
         (--group-by (funcall f it))
         ;; First item of the result is the input passed
         (cons data))))

(distribute (-iterate (lambda (_) (random 10)) (random 10) 20)
            (list (lambda (x)
                    (zerop (mod x 4)))
                  (lambda (x)
                    (zerop (mod x 2)))))

((8 2 7 3 6 7 3 1 6 2 1 3 0 7 5 7 8 1 2 0)
 ((closure (t) (x) (= 0 (mod x 4))) 8 0 8 0)
 ((closure (t) (x) (= 0 (mod x 2))) 2 6 6 2 2)
 (nil 7 3 7 3 1 1 3 7 5 7 1))

(distribute (-iterate (lambda (_) (random 10)) (random 10) 20)
            (list 'cl-oddp
                  'cl-evenp))

((0 0 5 0 3 7 4 3 2 7 9 0 3 5 8 8 6 8 3 2)
 (cl-evenp 0 0 0 4 2 0 8 8 6 8 2)
 (cl-oddp 5 3 7 3 7 9 3 5 3))

(distribute '(:hi "ho" "ho" :he 2.5)
            '(floatp
              keywordp
              stringp)
            )

((:hi "ho" "ho" :he 2.5)
 (keywordp :hi :he)
 (stringp "ho" "ho")
 (floatp 2.5))

[emacksnotes]

See also collect-cond! a variant of -separate and -separate-multi

The following thing was implicit in the discussion, but was never openly stated: The resulting subsets are disjoint, and an object can belong to one and only one bin.

There could be cases where we may want a single object to belong to multiple bins.

[emacksnotes]

See also collect-cond! a variant of -separate and -separate-multi

The following thing was implicit in the discussion, but was never openly stated: The resulting subsets are disjoint, and an object can belong to one and only one bin.

There could be cases where we may want a single object to belong to multiple bins.

If you look at the examples I cited above, you will see that the bin-function annotates a list of objects ('annotate' here is used in the sense of -annotate). Or viewed other way you are slapping tags--as in Org tags or Twitter tags--to objects. So, an obect can have multiple tags.

The word tag (in the sense of org tag) also make sense because the original request was raised by the org-superagenda and org-ql author ... So, instead of -distribute or whatever other names have been proposed choose a name that reflects what the operation here is; and the operation is "Annotate and Group by Annotation" (if you squint a bit)