Has anyone been seeing occasional coredumps in hv.c:hsplit()? I have a program here that is giving me this in a patchily reproducible way, but refuses to do the same when run over a debugging perl. The program also uses a couple of locally developed XS libraries, so I may be getting some random location overwritten, but it seems odd in that case that it always cores at the same spot. Running under Solaris 2.7. Thanks in advance for any clues, Hugo