Re: [Patch] Oracle on Mac OS X with SBCL and OpenMCL

Kevin Rosenberg wrote:

> I've committed your patch as is. I haven't had the time to work on
> fixing the issue with UFFI and OpenMCL. I've uploaded version 2.11.15
> with the changes.

Cool!  I took another look at the DEREF-POINTER definition for OpenMCL,
and I think the use of DEFSETF is the culprit as it is delaying the
evaluation of the :POINTER-VOID argument by rebinding it to a gensym.

? (get-setf-expansion '(uffi:deref-pointer foo :pointer-void))
(#:G136 #:G137)
(FOO :POINTER-VOID)
(#:G135)
(UFFI::DEREF-POINTER-SET #:G136 #:G137 #:G135)
(UFFI:DEREF-POINTER #:G136 #:G137)

Removing the conditionalized definition of DEREF-POINTER-SET and the 
corresponding DEFSETF appears to fix the problem (unless I'm missing
a subtle reason why MCL needs DEFSETF unlike the other Lisps...)

> In this release, the Oracle passes 216 out of 218 tests on
> Allegro/Linux/x86.

James