714218 - Move element/property/special-specific handling for a few classes out of getGeneric methods into specific get* methods

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[Jeff Walden [:Waldo]]

Attachment: Patch

Comment 1

[Brian Hackett [Laid off!]]

Comment on attachment 584904 [details] [diff] [review]
Patch

Review of attachment 584904 [details] [diff] [review]:
-----------------------------------------------------------------

::: js/src/jsarray.cpp
@@ +871,1 @@
>  }

Will the getGeneric hooks be going away entirely soon?

Comment 2

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

Comment on attachment 584904 [details] [diff] [review]
Patch

> array_getSpecial(JSContext *cx, JSObject *obj, JSObject *receiver, SpecialId sid, Value *vp)
> {
>-    return array_getGeneric(cx, obj, receiver, SPECIALID_TO_JSID(sid), vp);
>+    if (obj->isDenseArray()) {
>+        JSObject *proto = obj->getProto();
>+        if (!proto) {
>+            vp->setUndefined();
>+            return true;
>+        }
>+    }

Perhaps rewrite this as

if (obj->isDenseArray() && !obj->getProto()) {
    vp->setUndefined();
    return true;
}

?

Comment 3

[Jeff Walden [:Waldo]]

All the generic hooks will be going away "soon".  I'm not sure exactly what will replace them.  v8 just has all callsites convert to string, then fork on index-or-name at that point.  With our special junk, we can't do that so tersely.  It's an ugly problem I plan to think about today and hopefully come to some idea of a solution.

Comment 4

[Jeff Walden [:Waldo]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/be81e5f7850f

Comment 5

[Marco Bonardo [:mak]]

https://hg.mozilla.org/mozilla-central/rev/be81e5f7850f