On Tue, 3 Mar 2026 13:58:10 GMT, Alan Bateman <[email protected]> wrote:
>> Staging this as a Draft PR for now (FYI: @AlanBateman @DougLea)
>
> src/java.base/share/classes/java/lang/VirtualThread.java line 337:
>
>> 335: Continuation.pin();
>> 336: try {
>> 337: scheduler.execute(runContinuation);
>
> You can leave out the changes to VirtualThread, as we did for JDK 26
> ([pull/29187](https://github.com/openjdk/jdk/pull/29187)). This will ensure
> we use execute(ForkJoinPool) consistently rather than a max that includes
> execute(Runnable).
I did a `git revert` so it automatically backed out all changes for that
commit, I've now pushed a re-add of the VirtualThread-specific changes.
> src/java.base/share/classes/java/util/concurrent/ForkJoinPool.java line 3317:
>
>> 3315: if ((config & PRESET_SIZE) != 0)
>> 3316: throw new UnsupportedOperationException("Cannot override
>> System property");
>> 3317: return getAndSetParallelism(size);
>
> For the JDK 26 backout we decided to not revert the change to setParallelism
> as it's useful to trigger activation when increasing parallelism.
Nice catch, I'll re-apply that one as well.
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/30023#discussion_r2879054663
PR Review Comment: https://git.openjdk.org/jdk/pull/30023#discussion_r2883085810
