RFR(S): 8215890: [TESTBUG] jdk/jfr/api/recording/event/TestPeriod.java fails with "period delta too short"

[mikhailo.seledtsov at oracle.com]

Based on my earlier discussion with Erik and Markus I am withdrawing 
this change.

The test is inherently unstable. The recommended short-term action is to 
problem list the test. In longer term we should change the test to not 
rely on timing of events, as these can vary due to multiple factors, 
such as GC pause, compilation or other non-deterministic effects.


Thank you,

Misha


On 1/31/19 2:54 PM, mikhailo.seledtsov at oracle.com wrote:
> Please review this small change. The root cause of the bug was the 
> assumption that the events are ordered by
> time in the recording, which is not always true. There is no guarantee 
> of that in current design. The fix is to order the events of interest 
> by time before calculating the deltas.
> Also, while running preliminary testing I have noticed that on 
> occasion the test will time out. The root cause
> is that a test case iteration did not have enough time to record 
> minimum number of events. I have added a bit of
> extra time intended to fix this problem.
>
>     JBS: https://bugs.openjdk.java.net/browse/JDK-8215890
>     Webrev: http://cr.openjdk.java.net/~mseledtsov/8215890.01/
>     Testing:
>         Ran this test multiple times on Linux-x64, Mac, Win-x64, Sol-Spc
>            20 times on each platform - All PASS
>
> Thank you,
> Misha
>