Wed, 21 May 1997 00:37:29 -0400 (EDT)

JH> Is a possible invocation of set-version-limit, but as '(a . b) will 
JH> probably soon be garbage it's hardly sensible.  Well, if (set-version-limit)
JH> returned it's argument, then you could do a setf on the result, thus 
JH> changing '(a . b), but this is a contrived example.  I still think it
JH> doesn't make much sense on having versioning on something that can't 
JH> be referred to by name.

Anything _can_ be referred to by name.

(setq *happy-name* (cadar *sad-name*))

Unless you just mean versioning on symbols.

JH> The exception would be that possibility for reversible execution that you
JH> mentioned.  I guess in those limited contexts you might want versioning on
JH> everything.

I'm not sure why you want this extra lowlevelness, wouldn't a transluscent
versioning container do the trick?

Just forwards all method calls onto the most recent version, unless
you use its versioning-api?

Maybe I've missed the point.