RFR(S) 8194831 - [TESTBUG] Test runtime/CommandLine/OptionsValidation/TestOptionsWithRanges.java misleading error message
gerard.ziemski at oracle.com
Wed Jan 17 16:41:12 UTC 2018
> On Jan 16, 2018, at 4:07 PM, David Holmes <david.holmes at oracle.com> wrote:
> Hi Gerard,
> This looks okay to me.
> On 17/01/2018 7:50 AM, Gerard Ziemski wrote:
>> Hi all,
>> Please review the following simple fix for improving robustness of error reporting of TestOptionsWithRanges jtreg test:
>> There are 2 simple things we can do here:
>> - #1 Do not show the explicit GC (that the user set) when printing a failure of the test that was processing a runtime flag incompatible with that GC (and instead it implicitly used a different GC).
>> For example: a user might have started TestOptionsWithRanges test by passing in “UseG1GC”, but when the test processes a flag like “CMSSamplingGrain”, it will actually use “UseConcMarkSweepGC” and ignore “UseG1GC", but in the error output it will still show both “UseG1GC” and “UseConcMarkSweepGC”, which is confusing when we try to reproduce using these flags.
>> - #2 Show an error exit code both as integer (current behavior) and also as hexadecimal, which might help to pint point the actual cause on platforms like Windows
>> Tested locally and via Mach5.
More information about the hotspot-runtime-dev