> El 7 dic. 2021, a las 13:33, Manolo <manol...@gmail.com> escribió:
>
> I would sggest you run your program under a debugger and make sure the pointer you transmit
> as 3rg argument to Fl::repeat_timeout() remains valid until the moment when the timeout runs.
>

I debugged the program with printfs and found out that the dangling pointer is not the one as parameter to repeat_timeout (which is "this" of the class), but the current_timer->timer one. It seems there’s a running issue with the list of mac_timers and their timing to be ready to be called on repeat_timeout, leading them to be not initialized. I am unsure of this Fl_Cocoa_Screen_Driver.cxx code as it seems to be running on one thread but do things that several threads would do.


void Fl_Cocoa_Screen_Driver::repeat_timeout(double time, Fl_Timeout_Handler cb, void* data)
{
if (!current_timer) {
add_timeout(time, cb, data);
return;
}

//////////////////////// ADDED and fixes my crashes, but I consider this a hack and not a true solution.
if ( current_timer->timer == nullptr ) {
return;
}