SEL function may have the problem of precision loss

[chen4s666]

Hello.When variables of different types are used in SEL functions, the accuracy of variables may be lost.
For example, when I enter a variable B of type DINT and another variable C of type REAL. The converted IR code first converts the variable B to type float. When I need the variable B, it convert the float variable B to DINT and then assign variable D. If the DINT value is outside the precision range of float, precision loss occur.
If I assign the variable B to 16777217, the variable D will get 16777216.
My ST code is as follows:

PROGRAM main
    VAR
		a:BOOL;
		b:DINT:=16777217;
		c:REAL;
		d:DINT;
    END_VAR
	d:=SEL(a,b,c);
END_PROGRAM

It will generate the ir code as follows:

define void @main(%main* %0) section "fn-main:v" {
entry:
  %a = getelementptr inbounds %main, %main* %0, i32 0, i32 0
  %b = getelementptr inbounds %main, %main* %0, i32 0, i32 1
  %c = getelementptr inbounds %main, %main* %0, i32 0, i32 2
  %d = getelementptr inbounds %main, %main* %0, i32 0, i32 3
  %load_a = load i8, i8* %a, align 1
  %1 = icmp ne i8 %load_a, 0
  %load_b = load i32, i32* %b, align 4
  %2 = sitofp i32 %load_b to float
  %load_c = load float, float* %c, align 4
  %3 = select i1 %1, float %load_c, float %2
  %4 = fptosi float %3 to i32
  store i32 %4, i32* %d, align 4
  ret void
}

[chen4s666]

I think there may be other ways to handle it, such as the following c code.

if(a){
	d = c;
}
else{
	d = b;
}