Nick Craig-Wood <nick at craig-wood.com> writes: > It could be made to work I'm sure by getting the interpreter to check > for timeouts every few hundred bytecodes (like it does for thread > switching). Is there some reason not to use sigalarm for this?