haven't we identified and *fixed* all the tests which use j.t.l.Platform class and their own security policy (as part of 8210039 or other testlibrary refactoring)? in other words, all the tests which use j.t.l.Platform must have a policy file which grants Platform java.util.PropertyPermission read. so there should be no impact at all, unless there are tests which grants permissions to specific properties, and such policies are relatively easy to find and correct.
-- Igor > On Jul 19, 2019, at 9:59 AM, Alan Bateman <[email protected]> wrote: > > On 19/07/2019 17:55, Severin Gehwolf wrote: >> : >> My issue is with finding those tests :-/ If we know the set of *all* >> tests affected by the breakage we could do approach 2. Approach 1 (or >> 3) seems safer. > Any impact will be limited to tests using the test infrastructure but using > their own policy. I agree #1 or #3 is the safest. > > I just scanned the new webrev (02/webrev) and it looks good to me. > > -Alan
