RFR: 7903519 : jtreg/jtharness is missing features for basic crash testing

[andrlos]

On Wed, 13 Dec 2023 09:53:11 GMT, Jiří Vaněk <jvanek at openjdk.org> wrote:

>> this does not exactly add a crash testing option, however it allows users to define their own resultModifiers that they can force jtharness to use via advertising their resultModifier service in META-INF. I tested it with basic crashOnlyModificator and it works as intended - allows me to correctly skip and pass all the regular failures and only leaves me with crash tests as failures. Perfect for corner cases like JVM testing and such.
>> Also checked correct behavior in the logs, where the regular failure is still being reported and only the justification of the result modification is added at the bottom.
>
> src/com/sun/javatest/Script.java line 492:
> 
>> 490:      * @return status after modification (the proposedStatus by default)
>> 491:      */
>> 492:     private Status tryModifyResult(Status proposedStatus, TestDescription td){
> 
> this method should be synchronized, or the `resultModifier` should be safely  instantiated singleton. I would prefer the second. It would be good to die in initial stage, not late in runtime

I prefer the method to be synchronized to be honest. I am not seeing how to keep the change this simple with mixing singleton in it..

-------------

PR Review Comment: https://git.openjdk.org/jtharness/pull/57#discussion_r1425159215