Cost Optimisation

In order to reduce cost SQRL allows you to optimize AND and OR conditions. The function cost data can be generated
in your production environment and passed into the instance.

1
2
3
4
5
6
7

import { createInstance } from "sqrl";
const instance = createInstance({
    functionCost: {
        expensiveFunction: 10000,
        cheapFunction: 35
    },
})

Using this data SQRL will calculate a cost for each feature based on the recursive cost of everything it depends on.

For example, in the following code:

1
2
3

LET X := expensiveFunction();
LET Y := X + cheapFunction();
LET Z := Y + X;

• The cost of X will be 10,000 (using the functionCost) data above.
• The cost of Y will be 10,036 (the cost of the add function defaults to 1).
• The cost of Z will be 10,037. Depending twice on X does not affect its cost, but there is an additional addition that adds 1.

This data is then used when ordering AND or OR conditions. Since the order of the parameters does not affect the result in SQRL, we can safely re-order to make the code cheaper and/or faster to run.

For example, given the above code expensiveFunction() OR cheapFunction() will be re-ordered to calculate cheapFunction() first and only calculate the value of expensiveFunction() if it was not truthy.

Calculating costs in production

It is up to you to determine how much each function costs. One way to do this is to figure out how many requests/day a backend service processes and divide it by the total daily cost of running that service. Functions that are purely CPU-bound can be measured and use an average.