DaanVDH

[teal cove]

Hi I recommend reading: https://support.stripe.com/questions/rounding-rules-for-stripe-fees

[blazing cradle]

for example:

        mul := bi.Mul(bi, big.NewInt(14))
        div := bi.Div(mul, big.NewInt(1000))
        sum := div.Add(div, big.NewInt(25))

yeah i found that, but there's no example for when the fixed part is added,

for example you have a .25c fixed fee with a 1.4% percentual fee

you first calculate an amount lets say: 100 * 0.014 = 101.4

but do you round then or after adding the 25 cents to that amount

[teal cove]

Gotcha. I'm not sure offhand. I recommend going to support as the question will get escalated to someone who knows: https://support.stripe.com

[blazing cradle]

Thanks, will do!

[wicked condor]

does it matter, though? the fixed fee is never a fractional number of cents, so it shouldn't have any bearing on how the overall fee is rounded

[blazing cradle]

Yeah that's true, hadn't thought about it like that!

this should work right:

func (s service) calculateTransactionFees(sub primitive.Decimal128, method string) (int64, error) {
    bi, _, err := sub.BigInt()
    if err != nil {
        return 0, fmt.Errorf("could not calculate transaction fees")
    }
    switch method {
    case Card:
        mul := bi.Mul(bi, big.NewInt(14))
        div := bi.Div(mul, big.NewInt(1000))
        sum := div.Add(div, big.NewInt(25))
        return sum.Int64(), nil
    case iDeal:
        return 29, nil
    case BankContact:
        return 34, nil
    case GiroPay:
        mul := bi.Mul(bi, big.NewInt(14))
        div := bi.Div(mul, big.NewInt(1000))
        sum := div.Add(div, big.NewInt(25))
        return sum.Int64(), nil
    default:
        return 0, fmt.Errorf("missing payment method")
    }
}

[wicked condor]

I think the division there will round down, not round to nearest

separately, I don't think you actually need to use the big package for this; the amounts involved here are always going to be << int64 max val

[blazing cradle]

Yeah, that would probably be easier anyway.

Thanks for your help

[wicked condor]

anyhow - I don't think either the big or math packages supports quite exactly what you want re: rounding an integer to the nearest thousand

you could implement it by hand, I guess? it's not hard, though it feels wrong somehow

[blazing cradle]

No, also looked for it but couldn't find anything

[blazing cradle]

I'll probably just use int64

[wicked condor]

👍

[blazing cradle]

Thanks!