On 01/12/2014 18:54, Daniel Fuchs wrote:
on the test side would it be worth testing all public classes available
(e.g in rt.jar ?) to ensure that
Field.setAccessible works as expected and that we don't hit this issue
again ? It might be some
what of a heavy test for jtreg inclusion though.
It could be worth a try. But let's wait until JEP 220 is in.
I'm not sure why JEP 220 would be a factor for now. JDK 8u and 7u are
the production
JDKs for a while to come. Martin's testcase looks like a good approach.
I just
ran it through the rt.jar classes and had it test for
"f.setAccessible(true);" on all
getDeclaredFields in each class. Run time shouldn't be an issue I think
: (6-7 seconds)
class load: 19624 classes, 5162.67 ms total time, 0.2631 ms per class
getMethods: 19624 classes, 1076.00 ms total time, 0.0548 ms per class
regards,
Sean.
