Bug 186991 – Different output compared with v8 and spidermonkey

• Component: JavaScriptCore
• Status: RESOLVED DUPLICATE

Hello,
The following code behaves incorrectly (inconsistent with the standard and other engines).

arr0 = [ 1, 2, 3 ];                                                                                                                 
arr1 = [ 4, 5, 6 ];  
handler1 = { 
    get: function (oTarget, sKey) { 
        print('arg ' + '1' + ':get ' + sKey.toString()); 
        if (sKey.toString() == 'Symbol(Symbol.isConcatSpreadable)') { 
            arr1['length'] = 4294967294;
        }   
        if (sKey.toString() == 'length') { 
            arr1['length'] = 10; 
        }   
        if (Number(sKey.toString()) != NaN) { ; } 
        return Reflect.get(oTarget, sKey); 
    },  
    has: function (oTarget, sKey) { 
        print('arg ' + '1' + ':has ' + sKey.toString()); 
        if (Number(sKey.toString()) != NaN) { 
            return Symbol.search;
        }   
        return Reflect.has(oTarget, sKey);
     }   
}; 
var proxy1 = new Proxy(arr1, handler1); 
func = Array.prototype.concat.bind(arr0, proxy1); 
arr2 = func(); 
print(arr2);

The output is:
arg 1:get Symbol(Symbol.isConcatSpreadable)
arg 1:get Symbol(Symbol.toPrimitive)
arg 1:get toString
arg 1:get join
arg 1:get length
arg 1:get 0
arg 1:get 1
arg 1:get 2
arg 1:get 3
arg 1:get 4
arg 1:get 5
arg 1:get 6
arg 1:get 7
arg 1:get 8
arg 1:get 9
1,2,3,4,5,6,,,,,,,

However, it should be:
arg 1:get Symbol(Symbol.isConcatSpreadable)
arg 1:get length
arg 1:has 0
arg 1:get 0
arg 1:has 1
arg 1:get 1
arg 1:has 2
arg 1:get 2
arg 1:has 3
arg 1:get 3
arg 1:has 4
arg 1:get 4
arg 1:has 5
arg 1:get 5
arg 1:has 6
arg 1:get 6
arg 1:has 7
arg 1:get 7
arg 1:has 8
arg 1:get 8
arg 1:has 9
arg 1:get 9
1,2,3,4,5,6,,,,,,,

BT group
2018.6.25

[Gary Kwong [:gkw] [:nth10sd]]

I re-ran the testcase in comment #0 with git rev eb42a8967d53ebb95bd59b6d89662ac7fdf95a8b and it seems to show the intended output. It shows the wrong output with a build from mid-2017, or rev 011c994d52cc30bdec69aebed8ec1a025966b34a.

I then ran a bisection using robobisect, and it seems to point to the following commit as a possible fix to this bug:


commit e580b9a5fa9b5013b749e72f1c10946cb14af9aa
Author: caitp
Date:   Tue May 29 16:56:29 2018 +0000

    [JSC] Fix Array.prototype.concat fast case when single argument is Proxy
    https://bugs.webkit.org/show_bug.cgi?id=184267
    
    Reviewed by Saam Barati.
    
    </snip>
    
    git-svn-id: https://svn.webkit.org/repository/webkit/trunk@232261 268f45cc-cd09-0410-ab3c-d52691b4dbfc


Is bug 184267 a likely fix?

[Alexey Shvayka]

(In reply to Gary Kwong [:gkw] [:nth10sd] from comment #1)
> Is bug 184267 a likely fix?

Thank you for tracking it down, Gary!

*** This bug has been marked as a duplicate of bug 184267 ***