[gjs] ByteArray: don't use a resolve function

[Giovanni Campagna <gcampagna src gnome org>]

commit b1ed57ac2cac1bbbf5dba5c2c9b8f2b0850b3c42
Author: Giovanni Campagna <gcampagna src gnome org>
Date:   Fri May 10 18:28:13 2013 +0200

    ByteArray: don't use a resolve function
    
    It would do nothing anyway, because it was doing priv_from_js
    with the wrong arguments, seeing always NULL and returning TRUE.
    We don't need to actually define properties, we can just hook
    get and set.
    
    Also, we can prevent setting the JS stored value by storing
    undefined in the setter.
    
    https://bugzilla.gnome.org/show_bug.cgi?id=700337

 gjs/byteArray.c |   75 +++----------------------------------------------------
 1 files changed, 4 insertions(+), 71 deletions(-)
---
diff --git a/gjs/byteArray.c b/gjs/byteArray.c
index f8d189c..4597d68 100644
--- a/gjs/byteArray.c
+++ b/gjs/byteArray.c
@@ -49,11 +49,6 @@ static JSBool byte_array_set_prop      (JSContext    *context,
                                         jsid         *id,
                                         JSBool        strict,
                                         jsval        *value_p);
-static JSBool byte_array_new_resolve   (JSContext    *context,
-                                        JSObject    **obj,
-                                        jsid         *id,
-                                        unsigned      flags,
-                                        JSObject    **objp);
 GJS_NATIVE_CONSTRUCTOR_DECLARE(byte_array);
 static void   byte_array_finalize      (JSFreeOp     *fop,
                                         JSObject     *obj);
@@ -61,14 +56,13 @@ static void   byte_array_finalize      (JSFreeOp     *fop,
 
 static struct JSClass gjs_byte_array_class = {
     "ByteArray",
-    JSCLASS_HAS_PRIVATE |
-    JSCLASS_NEW_RESOLVE,
+    JSCLASS_HAS_PRIVATE,
     JS_PropertyStub,
     JS_PropertyStub,
     (JSPropertyOp)byte_array_get_prop,
     (JSStrictPropertyOp)byte_array_set_prop,
     JS_EnumerateStub,
-    (JSResolveOp) byte_array_new_resolve, /* cast due to new sig */
+    JS_ResolveStub,
     JS_ConvertStub,
     byte_array_finalize,
     NULL,
@@ -306,10 +300,8 @@ byte_array_set_index(JSContext         *context,
 
     g_array_index(priv->array, guint8, idx) = v;
 
-    /* we could have coerced a double or something, be sure
-     * *value_p is set to our actual set value
-     */
-    *value_p = INT_TO_JSVAL(v);
+    /* Stop JS from storing a copy of the value */
+    *value_p = JSVAL_VOID;
 
     return JS_TRUE;
 }
@@ -335,8 +327,6 @@ byte_array_set_prop(JSContext *context,
     if (!JS_IdToValue(context, *id, &id_value))
         return JS_FALSE;
 
-    byte_array_ensure_array(priv);
-
     /* First handle array indexing */
     if (JSVAL_IS_NUMBER(id_value)) {
         gsize idx;
@@ -348,63 +338,6 @@ byte_array_set_prop(JSContext *context,
 
     /* We don't special-case anything else for now */
 
-    /* FIXME: note that the prop will also have been set in JS in the
-     * usual hash table... this is pretty wasteful and bloated. But I
-     * don't know how to turn it off. The set property function
-     * is only a hook, not a replacement.
-     */
-    return JS_TRUE;
-}
-
-static JSBool
-byte_array_new_resolve(JSContext *context,
-                       JSObject **obj,
-                       jsid      *id,
-                       unsigned   flags,
-                       JSObject **objp)
-{
-    ByteArrayInstance *priv;
-    jsval id_val;
-
-    *objp = NULL;
-
-    priv = priv_from_js(context, obj);
-
-    if (priv == NULL)
-        return JS_TRUE; /* prototype, not an instance. */
-
-    if (!JS_IdToValue(context, *id, &id_val))
-        return JS_FALSE;
-
-    byte_array_ensure_array(priv);
-
-    if (JSVAL_IS_NUMBER(id_val)) {
-        gsize idx;
-        if (!gjs_value_to_gsize(context, id_val, &idx))
-            return JS_FALSE;
-        if (idx < priv->array->len) {
-            /* define the property - AAARGH. Best I can tell from
-             * reading the source, this is unavoidable...
-             * which means using "for each" or "for ... in" on byte
-             * arrays is a horrible, horrible idea. FIXME - but how?
-             *
-             * The issue is that spidermonkey only calls resolve,
-             * not get, as it iterates. So you can lazy-define
-             * a property but must define it.
-             */
-            if (!JS_DefinePropertyById(context,
-                                       *obj,
-                                       *id,
-                                       JSVAL_VOID,
-                                       byte_array_get_prop,
-                                       byte_array_set_prop,
-                                       JSPROP_ENUMERATE))
-                return JS_FALSE;
-
-            *objp = *obj;
-        }
-    }
-
     return JS_TRUE;
 }