Underflow error for float BigNumber

[pikeas]

Describe the bug
Small numbers like 1.2 cannot be turned in BigNumbers.

Reproduction steps

import {utils, BigNumber} from "ethers"
utils.parseUnits("1234").mul(1.2) // how I discovered the issue
BigNumber.from(1.2) // simpler case that still fails

Search Terms
BigNumber, float, underflow, fraction

After a bit of digging, it looks like BigNumber.from() should only be called with integers? If this is correct, it would be great if this could be added to the docs as it's a bit of a foot gun. Alternately, could from() call Number.isInteger() when it receives input of type number?

[ricmoo]

Make sure you read through all the notes in BigNumber too.

The BigNumber class is only for integers; for performing fixed-point maths, ethers provides the FixedNumber class.

Make sure when searching issues, you include closed issues, there is a lot on BigNumbers. Possibly too many to filter through. :)

[pikeas]

Make sure you read through all the notes in BigNumber too.

The BigNumber class is only for integers; for performing fixed-point maths, ethers provides the FixedNumber class.

Yes indeed, hence my suggestion! The docs mention the max safe integer value, but do not explicitly state that only integers can be used. The combination of misleading class name ("BigNumber"), docstring, docs, and error message ("underflow" instead of "only integers allowed") made debugging things much harder than they needed to be.