RFR(S) : 8219417 : bump jtreg requiredVersion to b14

David Holmes david.holmes at oracle.com
Fri Feb 22 03:53:16 UTC 2019


On 22/02/2019 8:07 am, Igor Ignatyev wrote:
> David,
> 
> I tend to agree w/ Alan here, I'd also prefer us to have required 
> version bumped in all test suites together as it decreases differences 
> in observed test behavior and simplifies test failure analyses. w/ that 
> being, test/jdk/sanity/client/ is a different beast, it's not a real 
> test suite, it's unclear if TEST.ROOT.template is ever used, and 
> requiredVersion is commented out in TEST.properties. I've filed 8219552 
> [1] and assigned it to Muneer(cc'ed), who runs and maintains these 
> tests, to deal w/ these files separately.

Okay. If that's the new policy that's fine.

> [2] is the webrev w/ TEST.ROOT files updated in all the test suites. 
> unless there are other comments, suggestions, objections, I'd like to 
> process w/ reviewing of this patch.
> 
> Side note: all (two) well-known publishers of jtreg builds have been 
> notified about the new version and have already created and published 
> binaries at their usual location.
> 
> [1] https://bugs.openjdk.java.net/browse/JDK-8219552
> [2] http://cr.openjdk.java.net/~iignatyev//8219417/webrev.01/index.html

Changes looks good.

Thanks,
David

> Thanks,
> -- Igor
> 
>> On Feb 20, 2019, at 1:10 AM, Alan Bateman <Alan.Bateman at oracle.com 
>> <mailto:Alan.Bateman at oracle.com>> wrote:
>>
>> On 20/02/2019 08:44, David Holmes wrote:
>>>
>>> On 20/02/2019 5:25 pm, Alan Bateman wrote:
>>>> I think it's important that everyone is using the same minium 
>>>> version of jtreg when running the tests. If some people are using 
>>>> b14 (by way of using jib) then they may end up pushing test changes 
>>>> that fail for the those of us that run jtreg directly. So I think it 
>>>> is better to just bump the required version everywhere, after 
>>>> socialization of course as it takes a bit of time to build or find a 
>>>> pre-built jtreg.
>>> I don't really care either way, but that's not what has been 
>>> happening to-date. Client area in particular still seems to be 
>>> requiring only 4.1.
>> It can be tricky to track down issues that stem from using different 
>> versions of the test runner so I think we should fix that and have 
>> both infrastructure and local testing use the same min version if 
>> possible. The tests for client-libs are under test/jdk where the 
>> TEST.ROOT has 4.2 b13 as the required version. There are a sanity 
>> tests that gatekeepers use to run SwingSet demos as sanity tests - 
>> Phil might be able to saw more about those and whether those tests 
>> care about the jtreg version or not.
>>
>> -Alan
> 


More information about the jdk-dev mailing list