On Mon, 24 Mar 2025 19:56:03 GMT, Kevin Rushforth <k...@openjdk.org> wrote:
> > We might want to consider imposing a limit on Linux and Windows, too. It > > could probably be done as a follow-up... > > Although doing it as part of this PR would be OK, too. Which would you prefer? And there it is, that open can of worms. I will put together a cross-platform version of this PR and do some testing. My first impulse was to do a focused PR that addressed the Mac-only issue reported by the customer. But if my testing is correct (I'll double-check) it looks like that same customer was within throwing distance of destabilizing the system on Windows also. And there is a testing advantage to having it be consistent across platforms. This PR makes a distinction between nested event loops started in Platform.runLater runnables and ones started while handling an input event (the former counts against the limit and the latter doesn't). A centralized solution will probably lose that distinction but I'm not sure that's important. Exceeding 240+ nested events loops is problematic either way. ------------- PR Comment: https://git.openjdk.org/jfx/pull/1741#issuecomment-2749384904
