P> The basic arguments I can see for replacing the lisp engine P> would be easing of support problems (separate from P> packaging/object systems, although that would obviously help P> things out a LOT), adding of features that are truly needed P> (threading, lexical scoping and packages to name three) to P> support improvements in current usage, and speedups. It would be neat to have something like the `fasl' files too... and done with mmap or something, so that three or four invocations of XEmacs running in one box would share those RAM pages.