UPGRADED-ARRAY-ELEMENT-TYPE: more thoroughly signal errors on unknown types.
[sbcl.git] / src / compiler / fixup.lisp
index 1a5e064..2e98849 100644 (file)
@@ -14,8 +14,8 @@
 
 ;;; a fixup of some kind
 (defstruct (fixup
-           (:constructor make-fixup (name flavor &optional offset))
-           (:copier nil))
+            (:constructor make-fixup (name flavor &optional offset))
+            (:copier nil))
   ;; the name and flavor of the fixup. The assembler makes no
   ;; assumptions about the contents of these fields; their semantics
   ;; are imposed by the dumper.
@@ -29,8 +29,8 @@
   offset)
 
 (defstruct (fixup-note
-            (:constructor make-fixup-note (kind fixup position))
-            (:copier nil))
+             (:constructor make-fixup-note (kind fixup position))
+             (:copier nil))
   kind
   fixup
   position)
 ;;; they find themselves trying to deal with a fixup.
 (defun note-fixup (segment kind fixup)
   (sb!assem:emit-back-patch segment
-                           0
-                           (lambda (segment posn)
-                             (declare (ignore segment))
-                             ;; Why use EMIT-BACK-PATCH to cause this PUSH to
-                             ;; be done later, instead of just doing it now?
-                             ;; I'm not sure. Perhaps there's some concern
-                             ;; that POSN isn't known accurately now? Perhaps
-                             ;; there's a desire for all fixing up to go
-                             ;; through EMIT-BACK-PATCH whether it needs to or
-                             ;; not? -- WHN 19990905
-                             #!+sb-show
-                             (when *show-fixups-being-pushed-p*
-                               (/show "PUSHING FIXUP" kind fixup posn))
-                             (push (make-fixup-note kind fixup posn) *fixup-notes*)))
+                            0
+                            (lambda (segment posn)
+                              (declare (ignore segment))
+                              ;; Why use EMIT-BACK-PATCH to cause this PUSH to
+                              ;; be done later, instead of just doing it now?
+                              ;; I'm not sure. Perhaps there's some concern
+                              ;; that POSN isn't known accurately now? Perhaps
+                              ;; there's a desire for all fixing up to go
+                              ;; through EMIT-BACK-PATCH whether it needs to or
+                              ;; not? -- WHN 19990905
+                              #!+sb-show
+                              (when *show-fixups-being-pushed-p*
+                                (/show "PUSHING FIXUP" kind fixup posn))
+                              (push (make-fixup-note kind fixup posn) *fixup-notes*)))
   (values))