On 25/04/2014 22:26, Brian Burkhalter wrote:
Hello,
Issue: https://bugs.openjdk.java.net/browse/JDK-8041683
Patch: http://cr.openjdk.java.net/~bpb/8041683/webrev.00/
It’s not really a solution but the patch changes BitLengthOverflow and
DoubleValueOverflow to catch the OOME and print a warning to STDERR instead of
failing outright. These OOMEs do not occur in JPRT runs with default
parameters, so at least this patch would allow the tests to run successfully
when there is enough memory but not to cause the test run to fail when there is
not enough memory. This is more test coverage than never running the tests.
This looks okay although I suspect these tests will will throw OOME most
of the time, as it's so common to limit the heap size (512MB is common,
I think is what the make files use for example).
-Alan.
