On Fri, 8 May 2026 03:38:23 GMT, Chris Plummer <[email protected]> wrote:

>> This update address two enhancements:
>>  - [8380075](https://bugs.openjdk.org/browse/JDK-8380075) [lworld] Update 
>> JVMTI spec to clarify that ObjectFree events are not sent for value objects
>>  - [8384049](https://bugs.openjdk.org/browse/JDK-8384049) [lworld] Minor 
>> GetLocalObject/GetLocalInstance spec wording update
>> 
>> The first one adds an `ObjectFree` event spec clarification that only 
>> identity objects are supported by this event.
>> The second one just slightly reformulates previously added spec 
>> clarifications for the JVMTI  `GetLocalObject` and `GetLocalInstance` 
>> functions, underlying that they work only when preview features are enabled.
>> 
>> The `ObjectFree` implementation does not need any matching update. Strong 
>> references are used for value objects instead of weak references in the 
>> `JvmtiTagMap`, so the value objects are kept alive while they are listed in 
>> the `JvmtiTagMap`. This means there are no `ObjectFree` events can be 
>> generated for value objects.
>> 
>> Testing:
>>  - No testing is needed for pure spec wording changes
>> 
>> 
>> ---------
>> - [x] I confirm that I make this contribution in accordance with the 
>> [OpenJDK Interim AI Policy](https://openjdk.org/legal/ai).
>
> src/hotspot/share/prims/jvmti.xml line 5892:
> 
>> 5890:               On return, points to the variable's value.
>> 5891:               When preview features are enabled, the frame's method is 
>> a non-identity object
>> 5892:               constructor and the requested local is the 
>> "<code>this</code>" object, the
> 
> I think this should stay as "value object", but I'll defer to @dansmithcode 
> and @dholmes-ora

Thanks. I was thinking about the same. Changed to "value object" now.

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

PR Review Comment: 
https://git.openjdk.org/valhalla/pull/2412#discussion_r3206412576

Reply via email to