2.js

var ds = {
    totalCapacity: 1000,
    usedCapacity: 0,
    items: {},
    updateCapacity: function(){
		this.usedCapacity = Object.values(this.items).reduce((t, {weight}) => t + weight, 0)
	},
    getItem: function (key) {
        return this.items[key] ? this.items[key] : -1
    },
    putItem: function (key, value, weight) {
        while (this.usedCapacity + weight > this.totalCapacity) {
            this.evictItem()
        }
        this.items[key] = {
            value: value,
            weight: weight,
            lastAccess: new Date()
        }
        this.updateCapacity()
    },
    evictItem: function () {
        let lowest = Number.POSITIVE_INFINITY
        let toEvict = ''
        for (let i in this.items) {
            itm = this.items[i]
            score = itm.weight / Math.log(new Date() - itm.lastAccess)
            if (score < lowest) {
                lowest = score
                toEvict = i
            }
        }
     	this.updateCapacity()
        delete this.items[toEvict]

    }
}

// computational effiency of getItem is O(n) as the reduce method in updateCapacity runs a function on each item in the list.
// I made this decision based on keeping the code clean when managing the capacity. The alternative I considered was looping through all the 
// items in the list to update capacity but this seemed like extra mental clutter despite probably being faster.

// computational effiency of putItem is O(n) as we have two (non nested) loops, meaning the number of iterations is
// roughly equal to the number of items we have stored