Symbol type is not filtered out of keys in types for fromPairs, keys, toPairs, and toPairsIn about types HOT 2 OPEN

Great catch! Symbols() are inherently non-enumerable. Do you have a solution in mind? If not I can look into fixing

from types.

While looking into this, I found out it's worse than just symbols. The current typing for toPairs includes any non-enumerable property in its output, including class methods!

Check out this simple instance of a class put through R.toPairs

Now let's put the same code into typescript/play

Simplified here, you get this

import * as R from 'ramda';

class Point {
  public x: number;
  public y: number;

  constructor(x: number, y: number) {
    this.x = x;
    this.y = y;
  }

  map(fn: (a: number) => number) {
    return new Point(fn(this.x), fn(this.y));
  }
}

const point = new Point(3, 2);

const asPairs = R.toPairs(point);
//    ^? ["x", number] | ["y", number] | ["map", (fn: (a: number) => number) => Point]
// but the value you get is `[["x", 3], ["y", 4]]`

What this means is we can never use keyof to determine what should be part of the output array of key/value pairs

Because you also would have map if you use a function to create and object literal:

const makePoint = (x, y) => ({
  x,
  y,
  map: fn => makePoint(fn(x), fn(y))
});

const point2 = makePoint(3, 2);
const asPairs2 = R.toPairs(point2);
//    ^? ["x", number] | ["y", number] | ["map", (fn: (a: number) => number) => Point]
// but the value you get is `[["x", 3], ["y", 4], ["map", null]]`

I've always thought that the way Object.entires was typed was just due to very old typescript, before it had the ability to be more specific

entries<T>(o: { [s: string]: T; } | ArrayLike<T>): [string, T][];

But I understand now that it's due to how keyof does not descriminate against non-enumerables and non-own-properties

I'm thinking that the types for toPairs and fromPairs just needs to be typed exactly like Object.entries and Object.fromEntries, respectively.

Same with Object.keys and R.keys, Object.values and R.values, etc

Shitty thing is this is going to be a breaking change

from types.