RFR: 8224974: Implement JEP 352

Alan Bateman Alan.Bateman at oracle.com
Mon Aug 19 14:36:20 UTC 2019


On 19/08/2019 12:29, Andrew Dinn wrote:
> :
> I renamed the test to MapSyncFail and modified it to run without
> restriction to a specific os or cpu.
>
> I also generalized it to run twice with a boolean arg which selects mode
> READ_ONLY_SYNC on the first run and READ_WRITE_SYNC on the second one.
>
> The logic of the test is now to expect
>
>   1) IOException if Unsafe.isWriteBackEnabled -> true
>   2) UnsupportedOperationException if Unsafe.isWriteBackEnabled -> false
>
> If the wrong exception or neither exception is thrown the test fails.
>
> Case 1 currently only applies for x86_64.
> Case 2 applies for all other architectures.
Thanks, this looks must better and avoids checking the architecture and 
exception messages.

I think webrev.12 looks good; I don't have any other comments.

-Alan


More information about the nio-dev mailing list