RFR: 8353835: Implement JEP 500: Prepare to Make Final Mean Final [v2]

[Alan Bateman]

On Fri, 26 Sep 2025 13:02:57 GMT, Volkan Yazici <vyazici at openjdk.org> wrote:

>> Alan Bateman has updated the pull request incrementally with two additional commits since the last revision:
>> 
>>  - Review feedback
>>  - Change ciField::initialize_from to use is_mutable_static_final, suggested by Vladimir Ivanov
>
> test/hotspot/jtreg/runtime/jni/mutateFinals/MutateFinalsTest.java line 109:
> 
>> 107:         String instanceMethod = list1.get(rand.nextInt(list1.size()));
>> 108:         String staticMethod = list2.get(rand.nextInt(list2.size()));
>> 109:         return Stream.of(instanceMethod, staticMethod);
> 
> * What is the rationale for choosing a random `pairOf(instanceMutator, classMutator)`?
> * Does "instance mutator gets followed by a class mutator" have any particular importance for the tests they are used?
> 
> I was looking at this argument supplier and thinking of `Collections.shuffle()` over a list containing all method names, preferably, multiple times.

The comment in the test description has ".. to avoid starting a child VM to test every mutation method". I'll see if I can improve this comment to make it clearer, or maybe your comment is that the method source needs this comment too?

The context here is that the test is launching a VM with -Xcheck:jni to check that it terminates with a "FATAL ERROR in native method".  Doing this for each of 18 cases would be expensive, can be 30+ seconds on macOS systems with debug builds. To keep the test execution down, the test choses one JNI function that attempts to mutate a final instance field, and one JNI function that attempts to mutate a static final field. The ensures the instance + static implementations are tested on each run. There is template expansion, and all functions will get exercised with enough runs.

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/25115#discussion_r2390255811