Removing a Double restriction: float support #175
Conversation
### What's done: - enabled the decoding of Float type
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
There was a problem hiding this comment.
ktlint found more than 10 potential problems in the proposed changes. Check the Files changed tab for more details.
| lineNo: Int, | ||
| limits: FloatingPointLimitsEnum, | ||
| conversion: (Double) -> T, | ||
| ): T = if (content in limits.min..limits.max) { |
There was a problem hiding this comment.
what's about suggestion to convert to lower type and then back and check that value is the same?
There was a problem hiding this comment.
@nulls , such suggestions valid, however looks a little bit confusing for integer types. But for floating-point types it can cause an error.
Floating-point arithmetics in Kotlin compiler is partially implemented using C++ interop. And in C++ standard it is explicitly told that floating point numbers cannot store all values exactly (not due to limits, but due to precision), so there can potentially be cases when doubleNumber.toFloat().toDouble() != doubleNumber due to loss of bits even when the number is inside Float bounds (-MAX_FLOAT..MAX_FLOAT).
So it is a big question for me if I really should rely on this hack and have potential (but with low probability) case when we corrupt the number and throw an exception (and with a higher complexity algorithm) instead of a simple solution to check of bounds.
As you know, I am not a risky guy, so I think that I would better use low risk option here (at least with floating point numbers). I am sure you would have done the same in this situation... 😄
But If you or @aSemy have very strong objections - I can rework it in the future...
### What's done: - enabled the decoding of Float type
What's done: