Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site fluke.UUCP
Path: utzoo!linus!decvax!microsoft!fluke!bhaskar
From: bhaskar@fluke.UUCP (K.S. Bhaskar)
Newsgroups: net.lang
Subject: Re: static links vs. displays
Message-ID: <773@vax2.fluke.UUCP>
Date: Thu, 20-Oct-83 17:17:01 EDT
Article-I.D.: vax2.773
Posted: Thu Oct 20 17:17:01 1983
Date-Received: Sat, 22-Oct-83 01:15:33 EDT
References: <5550@arizona.UUCP>
Organization: John Fluke Mfg. Co., Everett, Wash
Lines: 20

I found displays to be much easier to understand than static links (many years
ago, I had to work through a complicated example to finally convince myself
that they work).  As I see it, the only advantages of static links are (a) no
limit to the nesting of scope and (b) practicality.  The first is obvious.
For the second, most programs tend to access mainly locals and globals, most
computers have a limited number of registers to expend on displays and have
efficient access to locals on the stackframe, so static links tend to be quite
sufficient.

Personally, after reading an article (I think it was by David Hanson a couple
of years ago), I became a convert to the idea that nested scoping is harmful.
Besides, it doesn't solve any of the real problems in programming -- I find
issues like classes, dynamic storage allocation, run-time binding, etc. to be
far more important in programming than static scoping.

One advantage of displays over static links not mentioned earlier is in
debugging a compiler that generates code that causes a program to throw up all
over itself and its stack.

{allegra,lbl-csam,microsoft,sb1,uw-beaver}!fluke!bhaskar