RFR 8160034: The `this` value in the `with` is broken by the repetition of a function call

[Attila Szegedi]

> On 19 Jul 2016, at 18:20, Hannes Wallnöfer <hannes.wallnoefer at oracle.com> wrote:
> 
> Please review:
> 
> Webrev: http://cr.openjdk.java.net/~hannesw/8160034/webrev.00/
> Bug: https://bugs.openjdk.java.net/browse/JDK-8160034
> 
> The change in ScriptObject.megamorphicGet actually fixes the bug. It took me quite some time that it could be fixed so elegantly at this level. Before that, I tried to fix it on the Method handle level in WithObject. 

So… would you say this is a solution or a workaround? What I mean is that it is a nicely small change, but is it correct in all cases? Maybe it should be further constrained by both isMethod and isScope? (Not sure whether there’s a situation where it’s !isScope and would be incorrect, though)

> The cleanup in WithObject.lookup is a leftover from that. I noticed that it implements code paths for both named and unnamed operations, but WithObjects are always in scope, and scope operations are always named, so I added an assertion for that and removed the code handling unnamed operations.

Yep, that part is definitely nice, +1.

> 
> Finally, the change in NashornGuards is to exclude properties from WithObject expressions from the special scope guards, using the ordinary map guard instead. Previously, these guards caused properties in with statements to relink even if with objects had the same map.

+1 too.

Attila.

> 
> Thanks,
> Hannes