home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!mcsun!uknet!icdoc!usenet
- From: sjv@doc.ic.ac.uk
- Newsgroups: comp.lang.functional
- Subject: Re: Lazy languages considered questionable
- Message-ID: <1992Jul21.093354.16474@doc.ic.ac.uk>
- Date: 21 Jul 92 09:33:54 GMT
- References: <BrFpJ9.2Ky@news.cso.uiuc.edu> <1992Jul20.190137.10410@mnemosyne.cs.du.edu> <1992Jul20.213929.10686@m.cs.uiuc.edu>
- Sender: usenet@doc.ic.ac.uk
- Organization: Imperial College
- Lines: 28
- Nntp-Posting-Host: dse-mac-l1-83.doc.ic.ac.uk
-
- In article <1992Jul20.213929.10686@m.cs.uiuc.edu> morrison@dante.cs.uiuc.edu
- (Vance Morrison) writes:
- >...
- >That is the very point I was trying to make. Namely that when you
- >use 'lazy lists' or other structures that work only in lazy languages
- >you ARE worrying about order of evaluation. (Since you have to make
- >sure that your program terminates). Now granted, you have to worry
- >about order of evaluation in a strict language too, but since lazy
- >data structures are not allowed, this reason can be highly localized
- >(that is the termination properties of a piece of code are only
- >very loosely coupled to the inputs to that code). If infinite
- >structures are allowed, that reasoning is not longer has localized
- >(since figuring out the termination properties now depends strongly
- >on the input given).
- >...
-
- Please can we have some examples to illustrate these points?
-
-
- >The bottom line is that if you REALLY want to abstract order of
- >evaluation away, you have to use a STRONGLY NORMALIZING language
- >(they do exist, take type theory for example). Now order of evaluation
- >truely doesn't matter since any evaluation order terminates.
-
- On the face of it, evaluation order is related more closely to the
- Church-Rosser property than to strong normalization.
-
- Steve Vickers.
-