On Mon, 8 Aug 2022 02:27:15 GMT, David Holmes <dhol...@openjdk.org> wrote:
>> Stuart Marks has updated the pull request incrementally with one additional
>> commit since the last revision:
>>
>> More edits from Alex's suggestions.
>
> src/java.base/share/classes/java/lang/Runtime.java line 99:
>
>> 97: * <a
>> href="{@docRoot}/../specs/jni/invocation.html#destroyjavavm">DestroyJavaVM</a>.
>> 98: * This function is responsible for initiating the shutdown sequence
>> when the number of running
>> 99: * ({@linkplain Thread#isAlive() live}) non-daemon threads first drops
>> to zero. When the shutdown
>
> I think we need to align this more with the proposed text of JLS 12.8. It
> needs to be clearer that `DestroyJavaVM` must be used in conjunction with
> `CreateJavaVM` to get the desired termination behaviour. Otherwise, the
> proposed text allows for the possibility that the JVM will always terminate
> when the last non-daemon thread terminates, even if `DestroyJavaVM` has not
> been called. It needs to be clear that if the JVM is started by the JNI
> Invocation API, then it must also be terminated by the JNI invocationAPI to
> get the required behaviour.
@dholmes-ora OK I've updated the implementation note to try to make it quite
clear that the shutdown-at-zero-non-daemon-threads behavior is tied to
`DestroyJavaVM`. (This is good; I didn't know this before I embarked on this
exercise of updating these specs!) I'm not entirely sure of the state of JLS
12.8, but if I know Alex :-) it's unlikely to change much, and I think this
section is consistent with it.
-------------
PR: https://git.openjdk.org/jdk/pull/9437
