Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!wuarchive!zaphod.mps.ohio-state.edu!ncar!gatech!uflorida!haven!adm!cmcl2!kramden.acf.nyu.edu!brnstnd
From: brnstnd@kramden.acf.nyu.edu (Dan Bernstein)
Newsgroups: comp.lang.misc
Subject: Re: Array references cannot be made optimal
Message-ID: <13046:Nov1604:16:4390@kramden.acf.nyu.edu>
Date: 16 Nov 90 04:16:43 GMT
References: <27498@megaron.cs.arizona.edu> <6045:Nov1519:34:2290@kramden.acf.nyu.edu> <11900@life.ai.mit.edu>
Organization: IR
Lines: 26

In article <11900@life.ai.mit.edu> misha@just-right.ai.mit.edu (Mike Bolotski) writes:
> And I like the
> ever-so-subtle transition from optimal addition sequence to 
> pointer version of array code.  

See, this is what I mean about computer scientists. Anyone who bothers
to think about the problem will understand why, in a general
computation, converting array references to pointer references is the
same as finding an optimal addition chain. (Hint: In machine language,
how are array references implemented?)

By the way, why do you insist upon repeatedly misquoting the phrase
``optimal addition chain in a general computation''? Which word don't
you understand?

> And just how seriously do you expect your opinions to be taken if
> you haven't read the basic references about compiler construction
> or theory of computation?

I have read each of the AHU/ASU/etc books. Once. I find none of them
particularly useful as references, as everything they say is presented
more to my taste either in Knuth or in original research articles. I
certainly find it amusing that you consider the currently fashionable
elementary textbooks to be ``basic references.''

---Dan