float to char (or uint8 etc) invokes undefined behavior and produces random results (eg: `char(255.0)`)

Example 1

when true:
  proc c_printf*(frmt: cstring): cint {.importc: "printf", header: "<stdio.h>", varargs, discardable.}
  let num = 256.1
  let a1 = uint8(num)
  when defined case3c_sub:
    echo ord(uint8(num))
  let a2 = ord(uint8(num))
  c_printf("%d\n", a1.cint)
  c_printf("%d\n", a2.cint)

Current Output

nim r main 216 0

nim r -d:case3c_sub main 208 0

nim r -b:cpp main 0 0

nim r -d:danger main 0 0

Expected Output

option 1: CT error: uint8(float) should be illegal as it's meaningless and causes UB ditto with other things like char(float)
option 2: alternative is to issue a OverflowDefect when an overflow happens during the conversion

Example 2

when true:
  let num = 255.0
  echo ord(char(num))
  let z = char(num)
  echo (z, ord(z))

nim r main 0 ('\x98', 152)

nim r -b:cpp 255 ('\xFF', 255)

nim r -d:danger 0 ('\x00', 0)

Additional Information

1.5.1 73ce40aaf71196faa08072216bd826a94b5d40b1 root cause for https://github.com/nim-lang/Nim/issues/10421 which I've just closed in favor of this issue

links

https://frama-c.com/2013/10/09/Overflow-float-integer.html

nim-lang / Nim