On Mon, 3 Oct 2022 16:44:56 GMT, Mandy Chung <mch...@openjdk.org> wrote:
>> I agree that getting rid of the clone can help -- but since [JEP >> 416](https://openjdk.org/jeps/416) the generators modified here is mostly a >> fallback and the bulk of the use will use `MethodHandles` (unless you >> disable JEP 416 and fall back to the legacy impl). I was mostly curious if >> you had a startup or other benchmark running on mainline where the change >> you're doing here could be observed. > >> I agree that getting rid of the clone can help -- but since [JEP >> 416](https://openjdk.org/jeps/416) the generators modified here is mostly a >> fallback and the bulk of the use will use `MethodHandles` (unless you >> disable JEP 416 and fall back to the legacy impl). I was mostly curious if >> you had a startup or other benchmark running on mainline where the change >> you're doing here could be observed. > > @stsypanov like @cl4es said, I expect these dynamically generated methods > should only be generated during very early VM startup. Once the module > system is initialized, it will use method handles. > > How do you run the spring-based applications when you observe these methods > being called? While this change is trivial and no harm to include this > change, this code is planned to be removed in a future release. @mlchung I can put here a stack trace of the application invoking the code if you are interested ------------- PR: https://git.openjdk.org/jdk/pull/10526
