Decimal * Uint128 treats Uint128 as a raw Decimal value

[archseer]

Seems to be a "feature" so raw Uint128 values can be treated as decimals.

// NOTE: using a custom Decimal fork here since the constructor is normally private
   unimplemented!(
        "{} {} {} {}",
        gas_cost * Decimal::one(), // decimal * decimal
        gas_cost * Uint128::from(1u128), // decimal * u128
        gas_cost * Decimal::from_ratio(Uint128::from(1u128), 1u128), // decimal * decimal(u128 / 1)
        gas_cost * Decimal(Uint128::from(1u128)), // decimal * decimal(u128)
    );
//1 'not implemented: 0.00000000000338 0 0.00000000000338 0'

This is very surprising, since the Div implementation doesn't follow this behavior, and Mul doesn't this either.

Similar to #1155 I propose the Decimal() constructor to be made public. The Mul implementation should be fixed, then the old behavior can be more explicit via decimal_val * Decimal(uint_val).

[webmaster128]

Thanks for brining this up. I'm really unsure what the desired output type is.

There is also no generic and safe conversion

• Decimal -> Uint128: because information after the decimal point get lost
• Uint128 -> Decimal: because the range of Decimal is smaller since 18 digits are reserved for the data behind the decimal point

This makes it hard to say: Use this output type that can easily be converted.

[webmaster128]

Looking at https://stackoverflow.com/a/44552464/2013738, it feels like this should not be implemented at all.

[archseer]

This should be implemented. I'd expect decimal * uint128 to do a simple multiplication. Equivalent to decimal(decimal.0 * uint128) or decimal * decimal::from_ratio(uint128, 1) but more gas efficient than constructing a int/1 fraction then multiplying two fractions together.

[archseer]

i.e. I have $0.05 dollars (internally, stored as 5 cents, the lowest possible fraction, with a DECIMALS value of 2). I double my money (2 * 5) => $0.10

[ethanfrey]

Basically, there are two ways we could want.

200 * 50% = 100 (multiply my number by some decimal) - this is implemented as majority case
25% * 3 = 75% (multiply my percentage by a number to get a bigger percentage) - this is not typical, but should be possible.

Maybe this is included in #1186 ?

But if we have Decimal * Decimal -> Decimal that should be plenty. And you can convert your integer to a decimal with no loss before multiplying. Making it possible, even if not super simple syntax.

[webmaster128]

The mixed type arithmetic Decimal * Uint128 should be removed when we get the chance. Created a ticket for 2.0: #1485