RFR (S) 8140650: Method::is_accessor should cover getters and setters for all types

Coleen Phillimore coleen.phillimore at oracle.com
Thu Nov 5 17:34:35 UTC 2015


I am planning on reviewing this but I haven't had time.  And I'm trying 
to build Zero for a different reason.  I would be pretty unhappy if it 
broke Zero and would like it to not do that.  I haven't had a chance to 
read all of this yet.  I thought is_accessor was only for Zero so I 
guess I have to read more why this is getting changed.  I will try to 
get to it today.


On 11/5/15 12:24 PM, Aleksey Shipilev wrote:
> On 11/05/2015 07:59 PM, Vladimir Ivanov wrote:
>>>> If there's an RFE filed and Zero maintainers promise to fix it in a
>>>> prompt manner, I'm fine with what you proposed.
>>> Yeah, that's the plan.
>>>> Also, I had an idea: why don't you move is_simple_accessor() from Method
>>>> to some Zero-specific location? I don't see any value in keeping it in
>>>> Method.
>>> That's an interesting trick, but I think it will cause more harm: shared
>>> interpreter.cpp would have to know about that Zero-specific method, or
>>> otherwise intercept "too wide" accessor shape and narrow it down before
>>> it reaches Zero. Therefore, I believe is_simple_accessor is the lesser
>>> evil.
>> Ok. Looks good then.
> Thanks for taking time to review, Vladimir! I'll count that as the
> review from the compiler side.
> I think we still need a review from the runtime side. Folks? For the
> record, we are discussing this change:
>    http://cr.openjdk.java.net/~shade/8140650/webrev.01/
> The patch passes JPRT, RBT (hotspot_all), and the new regression test.
> Thanks,
> -Aleksey

More information about the hotspot-dev mailing list