Re: imag-part of a real and non-complex number should be exact zero

On Mon Jul 27, 2026 at 10:16 PM CEST, Peter McGoron via Chicken-users wrote=
:
> For example, `(imag-part 1.0)` should evaluate to `0`, not `0.0`.
>
> Rationale:
>
> 1. We should have `(eqv? z (make-rectangular (real-part z) (imag-part z))=
)`.
>
> 2. The number `1.0` is exactly on the real number line, not=20
> approximately on the real number line. This behavior is important for=20
> some numeric code near the branch cuts of functions.
>
> 3. `(eqv? 1.0 1.0+0.0i)` is #f, so the numbers should be non-`eqv?` in=20
> one of their parts.
>
> I do not believe this is required by the R5RS/R7RS, but it is allowed=20
> and is useful.

Makes sense. I'll submit a patch to chicken-hackers.

Thanks for the suggestion.


cheers,
felix