On Mon, 4 Dec 2023 20:51:20 GMT, Joe Darcy <da...@openjdk.org> wrote:
>> This slightly increases the wait for reference processing to complete to
>> accommodate long Xcomp compile times. Testing is underway.
>
> test/jdk/java/lang/Object/FinalizationOption.java line 92:
>
>> 90: System.gc();
>> 91: try {
>> 92: Thread.sleep(30L);
>
> I think it would be better to scale these kinds of sleep time bounds by the
> timeout factor used to run the jtreg test, which should already to scaled up
> for known-to-be-slower testing scenarios.
>
> There is a utility method to return this value in jdk/test/lib/Utils.java
I guess that sort of makes sense. The default timeout factor is 4 I think so I
guess I descale the existing total wait of 1000 ms so that the total wait is
now Utils.adjustTimeout(250)? Maybe others might offer an opinion? Since
these waits are a fixed timeout with no early exit I don't think we want them
to be too long but anything in 1-5 seconds really doesn't matter. That's part
of why I was just picking a static value.
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/16956#discussion_r1414525196
