Gunnar Evermann <ge204(a)eng.cam.ac.uk&gt; writes: No, these are unrelated. The only thing these two types of crashes had in common is that I experienced both wile using lookup.el. I can verify that the one with the dead processes is fixed by your patch which checks for dead processes. The other type of crash during using lookup.el was the one which crashed after `split-window', shortly after inserting Japanese. This was extremely annoying, as it was very frequent. Last Sunday I was finally able to reproduce it, but it was very late and I was to stupid to narrow it down small bit of sample code. After applying your patch it went away. I tried it on Linux and Solaris 2.5 and lookup.el hasn't crashed since. I don't notice any obvious slowdown nor any other obvious changes. Can I help to measure the slowdown somehow? What sort of thing should be measured? Mike