summaryrefslogtreecommitdiff
path: root/Lisp
diff options
context:
space:
mode:
Diffstat (limited to 'Lisp')
-rw-r--r--Lisp/method-invocation.lisp20
1 files changed, 17 insertions, 3 deletions
diff --git a/Lisp/method-invocation.lisp b/Lisp/method-invocation.lisp
index 1db5794..edbdc17 100644
--- a/Lisp/method-invocation.lisp
+++ b/Lisp/method-invocation.lisp
@@ -373,11 +373,25 @@ Returns: *result* --- the return value of the method invocation.
;; structs actually just give us pointers to them,
;; so we just put those pointers back into the
;; functions as arguments.
- (setf (argref :pointer i) arg))
+ ;;
+ ;; Note that the target type is a struct/union,
+ ;; not a pointer. This means that we actually
+ ;; have to pass a struct/union as an argument. We
+ ;; therefore ignore the memory space reserved for
+ ;; argument cells in the argument buffer and
+ ;; simply set the argument pointer directly.
+ (setf (cffi:mem-aref objc-arg-ptrs :pointer i)
+ arg))
((array)
;; This, too, might someday be ripped out and
- ;; replaced with something better.
- (setf (argref arg-c-type i) arg))
+ ;; replaced with something more flexible. For
+ ;; now, it's the same as for structs and unions.
+ ;; That's the nice thing about opaque C data
+ ;; structures: As a binding writer, we just pass
+ ;; them around without caring about their
+ ;; structure.
+ (setf (cffi:mem-aref objc-arg-ptrs :pointer i)
+ arg))
((id)
;; This case is actually interesting. We can do a
;; lot of automatic conversion between different