Brendan Eich [:brendan] Assignee
	Description • 15 years ago

See bug 471214 comment 68 et seq.

/be

	Brendan Eich [:brendan] Assignee
	Comment 1 • 15 years ago

Passes trace-tests -- at least trace-test/tests/basic/testMethodSet.js covers the empty object initialiser in a loop case.

/be

	Jason Orendorff [:jorendorff]
	Comment 2 • 15 years ago

Comment on attachment 420460 [details] [diff] [review]
fix

r=me.

In jsops.cpp, case JSOP_NEWINIT, the existing code locks obj around the call to  js_GetMutableScope. That is unnecessary, right?

	Brendan Eich [:brendan] Assignee
	Comment 3 • 15 years ago

(In reply to comment #2)
> (From update of attachment 420460 [details] [diff] [review])
> r=me.
> 
> In jsops.cpp, case JSOP_NEWINIT, the existing code locks obj around the call to
>  js_GetMutableScope. That is unnecessary, right?

Oh, because the object is newborn-isolated? Yeah, but look at js_GetMutableScope:

JSScope *
js_GetMutableScope(JSContext *cx, JSObject *obj)
{
    JSScope *scope, *newscope;
    JSClass *clasp;
    uint32 freeslot;

    scope = OBJ_SCOPE(obj);
    JS_ASSERT(JS_IS_SCOPE_LOCKED(cx, scope));

/be

	Brendan Eich [:brendan] Assignee
	Comment 4 • 15 years ago

http://hg.mozilla.org/tracemonkey/rev/91e31581b670

/be

	Robert Sayre
	Comment 5 • 15 years ago

http://hg.mozilla.org/mozilla-central/rev/91e31581b670