In the initial patch I signal SB-KERNEL::ARGUMENTS-OUT-OF-VALID-DOMAIN, this condition is derived from ARITHMETIC-ERROR. This is similar to Lispworks behavior. CLISP signals DIVIZION-BY-ZERO, which is a bit confusing. Is that possible to derive from ARITHMETIC-ERROR and REFERENCE-ERROR at the same time?


2010/4/30 Tobias C. Rittweiler <>
Roman Marynchak <> writes:

> Thanks for this explanation. Now I understand the behavior of EXPT in CLISP
> and Lispworks, and I will modify the patch for SBCL accordingly.

You can signal an appropriate subtype of sb-kernel::reference-error and
include '(:ansi-cl :function expt) as reference so future people bitten
by it will have a pointer into the right direction.


Sbcl-devel mailing list