setBigTimeout

Interestingly, this library seems to suffer from the opposite problem: where setTimeout can trigger earlier than expected, setBigTimeout can trigger never at all!

The problem is that when setBigTimeout is invoked with a floating-point number (and numbers are floating-point in JS by default), it keeps computing the time left till trigger in floating point. But FP numbers are weird:

    > 1e16 - 1 == 1e16
    true

At some point, they don't have enough precision to represent exact differences, so they start rounding, and this gets extremely more inaccurate as the value increases. For correct behavior, remainingDelay needs to be stored in BigInt.

Of course, this problem is mostly theoretical, as it starts happening at around 2^83 milliseconds, which doesn't even fit in a 64-bit time_t, and it's not like humanity will exist by then. But still!

I would go so far as to say entirely theoretical.