code review request: 7087428: move client tests out of jdk_misc
kelly.ohair at oracle.com
Wed Sep 7 15:43:44 UTC 2011
On Sep 7, 2011, at 4:07 AM, Alan Bateman wrote:
> Weijun Wang wrote:
>> root: http://cr.openjdk.java.net/~weijun/7087428/control/webrev.00/
>> jdk: http://cr.openjdk.java.net/~weijun/7087428/webrev.01/
>> I also update Makefile to put jdk_sound to its correct alphabetical position.
> The changes look okay to me. The jdk_misc batch is still an odd mix of tests but clearly fine grain splitting will lead to too many tiny batches.
> I don't know what to say about the jdk_swing and jdk_awt batches. It does seem a bit strange to have some 2d tests in the jdk_swing batch and others in the jdk_awt batch. I haven't seen any change-sets from folks in those areas adding or removing tests from the problem list so it's possible that they aren't using the make file to run the tests.
Not sure they run these tests much from the Makefile. Originally I had to corral them off because they belonged to a set of tests
that only worked with a DISPLAY setup on Solaris/Linux, and they seemed to poison the DISPLAY with strange settings.
That's when I created a unique VNC display on the JPRT driver machine for each user job in JPRT, starting the DISPLAY
fresh each time and killing it off when the job was done, so the poisoning was contained and didn't impact other user jobs.
Last time I ran them, they seemed to run ok. (and jdk_misc has some tests that need a DISPLAY too I think).
We should probably warn Amy (QE) about these tests. :^(
More information about the core-libs-dev