On 3/29/12 2:34 AM, Frederic Parain wrote:
Replying to both lists this time.
On 03/28/12 05:15 PM, Daniel D. Daugherty wrote:
>> Tested manually and with JPRT by running the jdk_management suite.
>> No new test has been added because the fix is in the HotSpot
repository,
>> and the GarbageCollectorMXBean notification unit tests are in the
>> jdk repository.
>
> You didn't mention whether you also ran with the test provided
> by the submitter.
Yes, I was not clear about the manual testing.
I've run the test provided by the submitter and used JConsole
to manually checks that the GcInfo attributes in the notification
were the same as the GcInfo attributes get from lastGcInfo.
> There is support for putting JavaTest/JTREG
> tests in hotspot/test/serviceability, but I'll have to agree
> that the rest of the related tests live in jdk/test. Are you
> planning to push a test to jdk repo sometime in the future?
Right now, I don't have such a plan. Writing a reliable automated
test for the jdk requires more work than the test case provided
in the CR. If you think such a test is required, I can fill a new
CR and start working on it.
I'm presuming that there is no existing test in the subsuite that
either caught this issue or could be easily and logically modified
to catch this issue. I'm always in favor of adding a test when you
fix a bug. However, you'll need to balance that general guidance
with the other things on your plate...
Dan
> Obviously, the usual coordination dance will have to come into
> play... waiting for the right hotspot to get promoted before
> pushing the test to the jdk repo... etc...
With experience, serviceability engineers become very good dancers
Thanks,
Fred
