On 17/08/2012 2:33 AM, Rob McKenna wrote:
Sounds good:
http://cr.openjdk.java.net/~robm/7191777/webrev.02/test/java/lang/ProcessBuilder/Basic.java.cdiff.html
<http://cr.openjdk.java.net/%7Erobm/7191777/webrev.02/test/java/lang/ProcessBuilder/Basic.java.cdiff.html>
I think we would want the non-timed version of waitFor. Otherwise the
default timeout will never kick in (unless the timing aspect of waitFor
is broken). Hmmm - can't test two things at once. See what Alan says. :)
David
-Rob
On 16/08/12 08:19, Alan Bateman wrote:
On 16/08/2012 03:18, David Holmes wrote:
Hi Rob,
On 16/08/2012 9:09 AM, Rob McKenna wrote:
Hi folks,
One of the tests from 4244896 failed once during nightly testing. It
isn't known how much of a delay will be necessary in order for it to
pass. In any case the tolerance can't really be loosened much more
without making the test meaningless so I've decided to remove it.
http://cr.openjdk.java.net/~robm/7191777/webrev.01/
<http://cr.openjdk.java.net/%7Erobm/7191777/webrev.01/>
Can we not leave the waitFor in place but simply not check how long
we waited? That way if it really takes "too long" we hit the default
test timeout.
That seems a good idea as it also exercises waitFor at around the time
that the process is terminating.
-Alan.
