This is rather bizarre. The problem is revision 43646. I have
double-checked by reverting just those two lines on my local latest
trunk, and the bisection is accurate - somehow revision 43646 is the
problem.
I see that the code it calls does some things with the packet display,
but the crash is happening even on the main page before loading a
capture, so I'm at a bit of a loss. Any ideas?
See: https://bugs.wireshark.org/bugzilla/show_bug.cgi?id=7494
(Comment #3 and the first part of comment #4).
> We probably should not call new_packet_list_recreate() when there's no
> PacketList.
I agree.
Calling new_packet_list_recreate() only if there's a PacketList
certainly gives
the previous (better) behavior. (What's a good way to do this ?)