1 files changed, 14 insertions, 3 deletions
diff --git a/upb/bindings/lua/upb.c b/upb/bindings/lua/upb.c
index 2b05938..87b39e6 100644
--- a/upb/bindings/lua/upb.c
+++ b/upb/bindings/lua/upb.c
@@ -1341,10 +1341,21 @@ static size_t align_up(size_t val, size_t align) {
   return val % align == 0 ? val : val + align - (val % align);
 }
 
-#define CHARPTR_AT(msg, ofs) ((char*)msg + sizeof(lupb_msg) + ofs)
-
 /* If we always read/write as a consistent type to each value, this shouldn't
- * violate aliasing. */
+ * violate aliasing.
+ *
+ * Note that the slightly prettier option of:
+ *
+ *   *(type*)(&msg->data[ofs])
+ *
+ * ...is potentially more questionable wrt the C standard and aliasing.
+ * Does the expression &msg->data[ofs] "access the stored value"?  If so,
+ * this would violate aliasing.  So instead we use the expression:
+ *
+ *   (char*)msg + sizeof(lupb_msg) + ofs
+ *
+ * ...which unambigiously is doing nothing but calculating a pointer address. */
+#define CHARPTR_AT(msg, ofs) ((char*)msg + sizeof(lupb_msg) + ofs)
 #define DEREF(msg, ofs, type) *(type*)CHARPTR_AT(msg, ofs)
 
 lupb_msg *lupb_msg_check(lua_State *L, int narg) {