segfault in mlt_frame_is_test_card (self=0x0) at mlt_frame.c:93
Brought to you by:
ddennedy,
lilo_booter
Crashed for me on two occasions, I think right after moving around clips in openshot.
This is on Debian unstable with
Openshot 1.4.3-1
melt 0.8.6-1 with a couple of cherry picks (the ones mentioned in https://sourceforge.net/p/mlt/bugs/181/)
Stack traces attached.
There is a race condition or a missing reference on the producer. The consumer is still running in thread 1, rendering frames. Meanwhile, in thread 9, it connects the consumer to a new producer, which is causing the previous producer to close while it is still being used in the rendering thread. The previous producer is actually going through full destruction since all references to it have been released. Should applications be expected to stop the consumer when changing the producer to which it is connected? Or, should applications be expected to hold a reference to the old producer until after connecting the consumer to a new producer? Or, should MLT add an additional reference to the producer somewhere?
Too old now, and no clear steps how to reproduce.