Well, I've finally managed to figure out why lilypond started coredumping on certain files, starting several versions ago. It was running out of stack. When this happens, Iten::find_broken_piece ends up being called with null pointers, which of course cause the segfault. It was mysterious, since the stack frame would be reset and I couldn't get a stack trace from gdb. Doubly mysterious, since the files wouldn't cause a coredump on the lilypond developers machines. The more I thought about it, the more it seemed like a resource problem, but there was plenty of memory and swap. Well, I'm not used to a unix with ulimit, I admit, so it took me a while to figure out that the stack argument needed to increased. Just FYI, should you have the same problem. sdb