NetNews Usenet Archive 1992 #16/NN_1992_16.iso/spool/comp/lang/function/975

Internet Message Format | 1992-07-28 | 2.0 KB

Path: sparky!uunet!elroy.jpl.nasa.gov!usc!wupost!waikato.ac.nz!comp.vuw.ac.nz!brian
Newsgroups: comp.lang.functional
Subject: Re: ** LAZY LANGUAGES FASTER?? **
Message-ID: <[e-mail]>
From: [e-mail] (Brian Boutel)
Date: Wed, 29 Jul 1992 04:04:13 GMT
Sender: [e-mail] (News Admin)
References: <[e-mail]> <[e-mail]>
 <[e-mail]> <[e-mail]> <[e-mail]>
Organization: Dept. of Comp. Sci., Victoria Uni. of Wellington, New Zealand.
Nntp-Posting-Host: circa.comp.vuw.ac.nz
Lines: 32


In article <[e-mail]>, [e-mail] (John Farrell) writes:
|> In <[e-mail]> [e-mail] (R.S. Nikhil) writes:
|> >In article <[e-mail]>, [e-mail] (John Farrell) writes:
|> >>   *Non-strict* evaluation saves time by not evaluating arguments which are not
|> >> needed.
|>
|> >Pardon my nit-picking, but this is not true.  I refer you to my earlier posting
|> >(last week?) in which I pointed out that evaluators CAN work on unnecessary
|> >arguments while preserving non-strictness.
|>
|>   My statement is true if you are implementing non-strict evaluation on a
|> sequential machine, in which case speculative evaluation can cause
|> non-termination unless you can prove that the argument being speculatively
|> evaluated was needed, and at the very least speculatively evaluating an
|> unneeded argument will waste time. Your nit-picking is pardoned :-).
|>
|>
|> John

To pick a further nit, you can, of course, simulate parallel evaluation on a
sequential machine, and so preserve non-strictness while working on unnecessary
arguments.

--brian


--
Internet: [e-mail]
Postal: Brian Boutel, Computer Science Dept, Victoria University of Wellington,
        PO Box 600, Wellington, New Zealand
Phone: [phone]   Fax: [phone]