Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!yale!cmcl2!kramden.acf.nyu.edu!brnstnd
From: brnstnd@kramden.acf.nyu.edu (Dan Bernstein)
Newsgroups: comp.lang.misc
Subject: Re: How to make a language downward-extensible?
Message-ID: <1047:Oct800:54:5490@kramden.acf.nyu.edu>
Date: 8 Oct 90 00:54:54 GMT
References: <18718:Oct120:03:0090@kramden.acf.nyu.edu> <7976@scolex.sco.COM> <13224@june.cs.washington.edu>
Organization: IR
Lines: 17

In article <13224@june.cs.washington.edu> pardo@june.cs.washington.edu (David Keppel) writes:
> * Anything on the list of approved alternatives is a valid
>   substitution.  You write `binsearch (...)' and the compiler
>   uses
> 	rule-based description
> 	quicksort
> 	any other magic way

No; that's exactly what I'm objecting to. I don't want to lose
portability just because I'm letting the compiler see several idioms for
the same computation. I also don't want to force every recipient of the
code to define binsearch or any other symbol just so they can compile.

You write out all the choices, explicitly, and the compiler hopes to
recognize one of them as something it can really optimize.

---Dan