[Sbcl-devel] Re: IMAGPART bug (x86/linux binaries)

[Paul D. <pau...@mo...>]

Gerd Moellmann wrote:

> Maybe there's an inconsistency in the standard?  The description of
> COMPLEX says
> 
>   If imagpart is not supplied, the imaginary part is a zero of the
>   same type as realpart; i.e., (coerce 0 (type-of realpart)) is
>   effectively used.
> 
> Wouldn't it be quite surprising if
> 
>   (imagpart -1.0)
>    => -0.0
> 
> and
> 
>   (imagpart (complex -1.0))
>    => 0.0


Perhaps.  I'll also note that the sentence from COMPLEX adds
another constraint to TYPE-OF and COERCE: the type returned
(for REALS) must contain zero, and must be an admissable second
argument to COERCE.

	Paul