Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!linus!decvax!duke!mcnc!ncsu!uvacs!mac
From: mac@uvacs.UUCP
Newsgroups: net.lang
Subject: {deep|shallow}{early|late}
Message-ID: <1086@uvacs.UUCP>
Date: Thu, 1-Dec-83 15:53:14 EST
Article-I.D.: uvacs.1086
Posted: Thu Dec  1 15:53:14 1983
Date-Received: Sun, 4-Dec-83 04:25:20 EST
Lines: 27



It's my impression that in the context of LISP there are two aspects to
binding: time and place.

Early and late binding are also called static and dynamic (or lexical and
?).  Lisp is traditionally late bound.  Scheme & related dialects are early
bound.

Deep and shallow binding are independent of binding times.  Deep bound
systems search a single binding stack for any variable (the search may go
deep).  Shallow binding systems have a separate stack for each variable and
need only look at the top (shallow).

According to this interpretation, deep/shallow binding is not visible in
the absence of concurrency.  Early/late binding makes a noticeable
difference.  Funargs want early binding.

It's true that there's confusion on this point.  The dragon book
("Principles of Compiler Design", Aho & Ullman), for one.  I was about to
complain about their incorrect use of the term deep to mean static binding,
when I discovered that the 1979 edition corrects this. "The Anatomy of
Lisp" is probably the final word.

Alex Colvin

 ARPA: mac.uvacs@csnet-relay CS: mac@virginia USE: ...uvacs!mac