Reach consensus on push/get/put order

[Izaakwltn]

Referring to Issue #938, I think we should try to reach consensus on the order of arguments in push, get, and put functions.

[stylewarning]

Thought: Currying the (quasi-)mutable collection seems wrong to me.

[stylewarning]

Elaboration: I think thinking about what you want to do with the curried versions informs us of what reads well. (nth n) is a good function to get the nth of something, while (ref dict) (a function which takes a key and returns a value) and its companion (get "5") (a function that takes a collection and returns a value) read nice in their respective cases.

[Izaakwltn]

I've also just had a moment of doubt, after recalling that the standard for Lisp list operations is arguments first:
(nth 0 ls)
(push obj objs)
(sort #'foo barlist)

(Of course, lisp arrays use the reverse order so that may be a weak point...)

[macrologist]

When and if loops become available, there would be less of a use-case for curried destructive functions.

At the present time, (iter:foreach (vector:push vec) (map g (iter:up-to some-number))) is kind of natural.

With loops, it becomes something like (perhaps)

(for i in (iter:up-to some-number) (vector:push vec (g i)))

[stylewarning]

I'd be happy if we just agreed on an accessor argument ordering to start with, and find a convention on the name. I think we should have two accessor names: GET and REF, with the key-or-index as the first and second arguments respectively in each.

[macrologist]

Works for me. Both terms are semantically broad enough to cover whatever exotic index types we happen to find ourselves passing.

[stylewarning]

Works for me. Both terms are semantically broad enough to cover whatever exotic index types we happen to find ourselves passing.

I also abhor using FLIP (or any other permuting combinator) for such common operations.

[macrologist]

I've been using flip - its fine about half the time; illegible the other half.