Newsgroups: comp.lang.lisp,comp.lang.scheme
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel-eecis!gatech!csulb.edu!hammer.uoregon.edu!news.uoregon.edu!cyclic.gsl.net!news.gsl.net!news.mathworks.com!news.maxwell.syr.edu!news.bc.net!torn!kwon!watserv3.uwaterloo.ca!undergrad.math.uwaterloo.ca!not-for-mail
From: papresco@csclub.uwaterloo.ca (Paul Prescod)
Subject: Re: Tractability of proving compiler correctness
Sender: news@undergrad.math.uwaterloo.ca (news spool owner)
Message-ID: <E6wn8y.LLq@undergrad.math.uwaterloo.ca>
Date: Wed, 12 Mar 1997 00:41:22 GMT
X-Newsposter: Pnews 4.0-test50 (13 Dec 96)
References: <5edfn1$83b@Masala.CC.UH.EDU> <hbaker-0603971513420001@10.0.2.1> <E6opD4.60p@undergrad.math.uwaterloo.ca> <33249420.46C9@ccs.neu.edu>
Nntp-Posting-Host: calum.csclub.uwaterloo.ca
Organization: University of Waterloo Computer Science Club
Lines: 21
Xref: glinda.oz.cs.cmu.edu comp.lang.lisp:25986 comp.lang.scheme:19085

In article <33249420.46C9@ccs.neu.edu>,
William D Clinger  <will@ccs.neu.edu> wrote:
>papresco@csclub.uwaterloo.ca (Paul Prescod) wrote:
>> Proving a compiler "correct" does not seem to be a tractable problem
>> (theoretically or practically)...
>
>For the record:  The algorithms that were used for code generation
>by the MacScheme compiler (at optimization level 0) and by the
>VLISP compiler were proved correct [1,2,3,4].  The compiler code
>that implemented those algorithms was not proved correct, but the
>code was a pretty straightforward transcription of the algorithms.

I was thinking more of an arbitrary compiler instead of one created
according to particular algorithms. Still, your experience is very
interesting: how much does designing according to provably correct
algorithms restrict your flexibility for optimization? Is it 
feasible for real-world code, or would the loss of optimization
render it infeasible?

 Paul Prescod

