RFR 8044047: Missing null pointer checks for streams
paul.sandoz at oracle.com
Tue Jul 15 11:44:37 UTC 2014
On Jul 15, 2014, at 12:38 PM, Daniel Fuchs <daniel.fuchs at oracle.com> wrote:
> On 7/15/14 9:35 AM, Paul Sandoz wrote:
> Hi Paul,
> This is not my area of expertise - but I haven't seen anything wrong.
> However - I applied your patch and ran the tests locally on my
> machine, and I saw that
> was intermittently failing in timeout (e.g. while I was compiling
> another JDK in another workspace).
> So I wonder whether something should be done for this test to reduce
> the risk of seeing it timing out on slow machines?
This test is not usually problematic in time outs AFAIK re. execution on the test infrastructure, usually other tests are problematic IIRC.
It runs under 30s on my laptop, more often < 15s for 182 tests (it's actually more because there is the cross-produt of streaming strategies not reflected in TestNG reporting).
If you use -Xcomp then it is a different matter. There is the timeout factor that can be used to adjust the time out when using arguments that result in slower execution or for slower machines.
I don't think we should be having to update tests that run under 30s (at least 4x of the default timeout) on a two year old laptop. Instead we should be taking a good long hard look at our test infrastructure and the way we execute (e.g. run each test method separately)
More information about the core-libs-dev