RFR for JDK-8028711: TEST_BUG: Tests should pass through VM options: corelibs tests
Alan Bateman
Alan.Bateman at oracle.com
Fri Feb 7 11:03:23 UTC 2014
On 07/02/2014 03:41, michael cui wrote:
> Hi Alan,
>
> Based on current discussions we have had and existing usages in our
> code base,
> I would like to propose that the fix should ONLY make sure that
> ${TESTVMOPTS} will be passed in when launching java from shell script.
>
> For other options such as ${TESTJAVAOPTS} ${TESTTOOLVMOPTS} and
> ${TESTJAVACOPTS},
> Since we are not very clear about needs to make them presents all the
> time when launching java or other java tools from shell, it is better
> to not make changes to current usages.
>
> Please let me know if you are ok with this approach.
If I understand correctly then you are proposing to address the
remaining issues where the vmoptions aren't passed through from shell
tests. That would be good. As I mentioned, there have been efforts on
this already but we either missed some tests or we didn't continue it
into newer tests. The other part is java tests that launch additional
VMs as we do have a few that don't pass the options through. That might
be a separate bug but would be good to address.
As regards TESTJAVAOPTS then I think it might be worth starting a
discussion on the jtreg list to see if there are opportunities to
consolidate the number of variables that tests have to deal with. I
don't have any objection to the original proposal, it's just that it
makes it obvious that there are too many options and too easy to get it
wrong.
-Alan.
More information about the core-libs-dev
mailing list