On Fri, 16 May 2025 19:03:21 GMT, Andy Goryachev <ango...@openjdk.org> wrote:
>> We could also just keep it with a fixed threshold, but remove the system >> property. > > It just seems unnecessary to add all this complexity. The first exception is > all we need, really - we can skip the 2nd, 3rd, ... This might rule out the > property. > > It might be unlikely that two or more completely unrelated failures would > happen in the field in a well-tested application [citation needed], I think > there is a little value in trying to add this complex logic per timer. A > static boolean might suffice. On the other hand, AnimationTimer is a public API that allows for arbitrary code execution, which always has the potential to fail for any number of reasons (think of calling an unstable API from the timer callback). We can’t rule that out entirely. This is only a fallback in any case, as a well-tested application shouldn’t throw exceptions from timer callbacks in the first place. ------------- PR Review Comment: https://git.openjdk.org/jfx/pull/1811#discussion_r2093564875
