Re: Down with weak lists!

BP> Why would you _not_ want these to behave in exactly the same way from lisp, BP> other than creation time and GC time? If they behave like a normal list, BP> experimenting with which is better for a specific application is incredibly BP> simple. There are no separate interfaces for weak hash tables (which _are_ BP> used in 3rd party lisp code). It's pretty much the same reason why keymaps aren't lists or vectors. Tying cache policy to some unknown GC policy is a bad idea. You just don't know what it's going to do. BP> Switching the glyph or face caching in Emacs/W3 to use weak lists BP> would be very simple the way it stands now, otherwise I'd have to BP> change many more places in the code. Well, I'm sure a suitable abstraction could easily generalize over both. -- Cheers =8-} Chipsy Friede, Völkerverständigung und überhaupt blabla