float32 arithmetics #2652

Description

Arithmetic operations on a float32 may produce numbers that are only representable as float64.

Repro code

This prints

2.7
2.7

but 2.7 is not a number that is representable as float32. On .NET, the same code prints

2.7
2.700000048f

Doing Math.fround(2.7) in the console returns 2.700000047683716, which corresponds to the .NET float32. Shouldn't all float32 operations, not just literals, be wrapped in an fround?

Related information

• Fable version: 3.4.10 in the REPL