Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!samsung!slug!zaphod.mps.ohio-state.edu!rpi!bu.edu!snorkelwacker!spdcc!esegue!compilers-sender
From: tli@%phakt.usc.edu (Tony Li)
Newsgroups: comp.compilers
Subject: Re: Unsafe Optimizations (WAS: Compiler Design in C How about it?)
Keywords: optimize, parallel
Message-ID: <1990Jun7.190315.4578@esegue.segue.boston.ma.us>
Date: 7 Jun 90 19:03:15 GMT
References: <1990Jun1.194941.5781@esegue.segue.boston.ma.us> <1990Jun4.212226.18389@esegue.segue.boston.ma.us> <1990Jun7.010349.2097@esegue.segue.boston.ma.us>
Sender: compilers-sender@esegue.segue.boston.ma.us
Reply-To: tli@%phakt.usc.edu (Tony Li)
Organization: University of Southern California, Los Angeles, CA
Lines: 24
Approved: compilers@esegue.segue.boston.ma.us

In article <1990Jun7.010349.2097@esegue.segue.boston.ma.us> larus@primost.cs.wisc.edu (James Larus) writes:

    Although parallel prefix may produce a different result, which is
    right and which is wrong? 
    
That depends.  If the semantics of the language specifies an order of
evaluation, then a "correct" compiler must implement that order of evaluation.


If your optimizer does something different then it is either a) broken, or b)
implementing a different semantics, which is a different language.  If you
claim that your compiler correctly implements language X and you choose option
b, then you mislead your customers.  You may claim that your compiler
implements extensions to language X, but if you do so, I would expect the
manual to describe the differences clearly.  For example, I would expect that
your optimizer switch would tell me in big, blinking red letters that it
introduces nonstandard semantics.

Tony Li - USC Computer Science Department
Internet: tli@usc.edu Uucp: usc!tli
-- 
Send compilers articles to compilers@esegue.segue.boston.ma.us
{spdcc | ima | lotus}!esegue.  Meta-mail to compilers-request@esegue.
Please send responses to the author of the message, not the poster.