[macOS w/ SDL2] Assertion failure sending mouse click events

[tarmac]

Ah, I believe the problem is that SDL itself overrides sendEvent: https://github.com/libsdl-org/SDL/blob/ ... nts.m#L101

If I understand correctly, the message to CEF would get sent in the call in line 101. But since I swizzled the sendEvent method in NSApplication, the `CefScopedSendingEvent` would affect only the method call in line 104. So, I believe I should be extending/swizzling SDLApplication(haven't been able to do this yet since it looks like it isn't exposed by SDL )

I'll try to figure my way around SDLApplication, but does my understanding of the problem make sense?

[tarmac]

You’ll likely need to start with a breakpoint in main() after the framework is loaded, or the file/line breakpoint in CEF/Chromium won’t resolve.

So I'm able to set a breakpoint, and the lldb output seems to indicate it found the correct method but the breakpoint doesn't get hit when I see the fatal error log message.

[magreenblatt]

I would expect that FATAL error to crash your application. Something strange is going on there.

If Cocoa_DispatchEvent is calling back into Chromium (outside of CefScopedSendingEvent as you describe above), that might explain why you are still having the issue.

[tarmac]

If Cocoa_DispatchEvent is calling back into Chromium (outside of CefScopedSendingEvent as you describe above), that might explain why you are still having the issue.

Huh, the SDL API lets you poll for events and then handle them. So, if I understand correctly, even with `[SDLApplication sendEvent]` swizzled, the handling of the SDL event would always be outside of CefScopedSendingEvent.

That is, the call to sendEvent(and thus CefScopedSendingEvent scope) seems to happen when calling SDL_PollEvent while CEF is sent the event after the call to SDL_PollEvent.

Perhaps the way forward then is to attempt setting/unsetting the `g_handling_send_event` boolean in a function that includes both the polling and event handling?