RFR(L): 8205044: [lworld] Interpreter and compiler support for acmp with value type operands (original) (raw)
John Rose john.r.rose at oracle.com
Thu Jul 19 06:09:23 UTC 2018
- Previous message (by thread): RFR(L): 8205044: [lworld] Interpreter and compiler support for acmp with value type operands
- Next message (by thread): RFR(L): 8205044: [lworld] Interpreter and compiler support for acmp with value type operands
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Sorry, it got lost in my inbox; thanks for the ping.
This looks very good to me; thanks for refactoring it that way.
And I think all of the node logic is now locally correct, regardless of what the other methods do.
— John
On Jul 16, 2018, at 5:21 AM, Tobias Hartmann <tobias.hartmann at oracle.com> wrote:
Hi John, On 13.07.2018 23:33, John Rose wrote: I think your methods that potentially return a constant result should be moved from Ideal to Value. That's not possible because these methods may return a new node or a constant. We would need to split the checks up and move them to Ideal or Value. I did this by re-factoring the code once more: http://cr.openjdk.java.net/~thartmann/8205044/webrev.04inc/ All tests pass. Thanks, Tobias
- Previous message (by thread): RFR(L): 8205044: [lworld] Interpreter and compiler support for acmp with value type operands
- Next message (by thread): RFR(L): 8205044: [lworld] Interpreter and compiler support for acmp with value type operands
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]