[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
- To: MOON at MIT-AI, BUG-LISPM at MIT-AI
- From: Daniel L. Weinreb <dlw at MIT-AI>
- Date: Thu ,2 Oct 80 13:17:00 EDT
- Sender: dlw at CADR8 at MIT-AI
I hate to say this (since it suggests doing something the hard way), but
if MM List Callers is going to know about tracing, then we are getting
into a little bit of the danger of Interlisp, where every advanced
feature knows about every other such feature. I think maybe we need a
general "advise" "protocol" by which any facility (the built-in Trace,
people's own Trace, metering) can attach stuff to a function definition,
using a standard protocol so that MM List Callers and suchlike can
ignore all advising and get at the real function in a standard way.
(Somebody stomp on me if things already work this way.) I do think letting
people add their own advising is important.