Recommended way to get a percentage?

[decafboba]

For uint256 specifically, let's say I have some uint256 with a value (ie, 131760045705399555255526021) and I want to take 25% of it in the most precise way possible, what would be the best way to do this via this library?

If this is a common use-case, on small request might be to have a perc function.

Thanks!

[PaulRBerg]

Hey @kenvogg, there are examples in the README.

In your case, you would write:

function takePercentage(uint256 x) external pure returns (uint256 result) {
  const percentage = .25e18; // 0.25 means 25%
  result = x.mul(percentage);
}

To be honest I don't see the value of a perc function. It'd be just syntactic sugar for mul.

[PaulRBerg]

This is a question for the evm-bn discussions channel. But, for the sake of helping you quickly, here's the answer:

const foo = toBn("25"); // first option
const bar = toBn("25", 18); // first option

By default, the toBn function applies 18 decimals to the input. Refer to the README of the evm-bn repo for more details.

Almost seems like values below "1" are impossible for TypeScript BN, is that correct?

No. toBn("1") would yield 1e18, which is 1.

[decafboba]

Hey @kenvogg, there are examples in the README.

In your case, you would write:

function takePercentage(uint256 x) external pure returns (uint256 result) {
  const percentage = .25e18; // 0.25 means 25%
  result = x.mul(percentage);
}

To be honest I don't see the value of a perc function. It'd be just syntactic sugar for mul.

If the percentage was not fixed (so cannot assume 25% in all cases), this way wouldn't work because floating points can only be const, right? What would be the recommended solution in that case?

[PaulRBerg]

I don't understand what you mean, PRBMath doesn't work with floating points, just fixed points.

Could you please restate your question?

[decafboba]

What I mean is, instead of the example takePercentage with fixed 25%, it would be:

function takePercentage(uint256 x, uint256 percentage) external pure returns (uint256 result) {
  result = x.mul(percentage);
}

And because a const can't be used, this wouldn't work as far as I know. So I know believe the next best way to get a percentage is:

function takePercentage(uint256 x, uint256 percentage) external pure returns (uint256 result) {
  result = x.mul(percentage).div(100);
}

This works if, for example, percentage is 25 to represent 25%. Some questions in regards to this though:

1. Does PRBMath provide any utility here or should we just do result = x * percentage / 100?
2. If we wanted to do a percentage lower than 1%, such as 0.1%, this wouldn't be possible with the current method. Is there some way PBRMath could help with getting some more precision?

Thanks in advance!

[PaulRBerg]

1. No, PRBMath provides no utility for percentages, you should just do x * percentage / 1e20 (not 100, because again, PRBMath treats 1e18 as equal to 1, PLEASE read the README).
2. It would be possible to do 0.1%. 0.1% is 1e17. Please read the README.