On 03/10/2012 22:44, Rob McKenna wrote:
Hi folks,
The only way I can see this test failing in this manner[*] is if we
destroy the process before we begin the read. That being the case I've
jacked up the sleep (giving the reader thread a little more time to
get cracking) and added a check to see if the threads stack has
entered a read call.
http://cr.openjdk.java.net/~robm/7152183/webrev.01/
<http://cr.openjdk.java.net/%7Erobm/7152183/webrev.01/>
Feedback greatly appreciated.
-Rob
[*] le trace:
So stack traces are masculine, I didn't know that.
I think your analysis is right, it's just that the sleep(10) is not
sufficient to ensure that the thread gets to the read method. Increasing
the sleep is probably sufficient. The hack to look at the stack trace
makes it more robust for really extreme cases, at the cost of potential
further maintenance in the event that the implementation changes. In any
case it's good to resolve this intermittent test failure.
-Alan
