Yes, this commit avoided the deadlock, because pthread_join() was not
called.
But, the program was messed up because of the call to the rogue
eina_semaphore_destroy(), so it was not a fix, just something that prevented
some symptoms of the problem to manifest in one particular way.
Eina does not h
Okay, but wasnt that commit fixing the deadlock you had in the join?
Further more, after evas_thread_exited is 1 only eina deals with this
thread, not evas_thread_renderer.c so not waiting for the thread there
seems to be fine IMO, but keeping the code smaller is probebly better.
Greetings
Mar
Hi,
I think this is just wrong.
It evas_thread_exited is True, it does not mean that the thread is actually
finished. eina_thread_join() should be called **unconditionally** in this
case.
It is specified that if the thread is terminated, then pthread_join()
(eina_thread's backend) will return i
