RFR: 8196657: [Testbug] TestOptionsWithRanges will fail for newly obsoleted flags that have not yet been removed

[Kim Barrett]

> On Feb 2, 2018, at 8:27 PM, David Holmes <david.holmes at oracle.com> wrote:
> 
> Updated webrev:
> 
> http://cr.openjdk.java.net/~dholmes/8196657/webrev.v1/

Looks good.

> 
> Thanks,
> David
> 
> On 2/02/2018 11:09 PM, David Holmes wrote:
>> Hi Kim,
>> On 2/02/2018 9:41 PM, Kim Barrett wrote:
>>>> On Feb 2, 2018, at 3:39 AM, David Holmes <david.holmes at oracle.com> wrote:
>>>> 
>>>> bug: https://bugs.openjdk.java.net/browse/JDK-8196657
>>>> webrev: http://cr.openjdk.java.net/~dholmes/8196657/webrev/
>>>> 
>>>> TestOptionsWithRanges uses PrintFlags to get the set of flags to test, and then proceeds to give them valid and invalid values.
>>>> 
>>>> When we bump the JDK version and a flag is now considered obsolete (but has not yet been removed from the code) the flag is ignored (other than printing a warning) and so the tests with invalid values do not generate errors, and so those tests fail.
>>>> 
>>>> The fix is to simply watch for "Ignoring option xxx" in the stderr output and report success.
>>>> 
>>>> Pushing to jdk/jdk.
>>>> 
>>>> Thanks,
>>>> David
>>> 
>>> Shouldn’t the fatal error detection case still be first?  As written, if an option is ignored but
>>> the VM then failed to start for some reason, that will be counted as a pass.  That doesn’t
>>> seem right.
>> Okay ... seems a fairly remote possibility given we're only doing "java -version" plus the flag setting under test, but yes theoretically possible.
>> I will switch the order.
>> Thanks,
>> David