Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 (Tek) 9/26/83; site tekecs.UUCP
Path: utzoo!linus!decvax!tektronix!tekgds!tekecs!patcl
From: patcl@tekecs.UUCP (Pat Clancy)
Newsgroups: net.lang.lisp
Subject: Re: shallow binding and deep binding
Message-ID: <3272@tekecs.UUCP>
Date: Tue, 29-Nov-83 14:25:24 EST
Article-I.D.: tekecs.3272
Posted: Tue Nov 29 14:25:24 1983
Date-Received: Thu, 1-Dec-83 04:33:45 EST
Organization: Tektronix, Wilsonville OR.
Lines: 41

The definitions of the terms "shallowing binding" and
"deep binding" which I learned are different from those
given by umcp-cs!liz. Namely: deep binding requires a tree
of saved environments, while shallow only requires a
stack. The difference has to do with the binding of free
variables, eg:

	(csetq foo(lambda(x)
		(lambda()(car x))
			))
	
	(csetq bar(lambda(x y)
		((foo y))
			))

(please excuse the probably weird usage, I learned Lisp
on a Univac 1100!)

Now the result of: (bar '(a b) '(c d))
has two possible values:

deep binding:	c
shallow binding: a

The free variable is the "x" inside the function returned
by "foo". With deep binding, the alist branch created when
foo is being evaluated is saved, including the entry for x,
and this is the x which is found on evaluation of the inner
(unnamed) function (ie, the inner function is passed a "funarg"
including a pointer to the saved environment). With shallow
binding, the x created during foo's evaluation is lost when
foo exits (no alist tree, only a stack), so the x created
during evaluation of bar is found.

I have seen both this, and umcp-cs!liz's definitions in
the literature, so I believe they are both correct but
incompatible usages of these terms.

Pat Clancy
Tektronix