Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!watmath!clyde!burl!ulysses!allegra!princeton!caip!lll-crg!seismo!mcvax!enea!sicsten!lhe From: lhe@sicsten.UUCP (Lars-Henrik Eriksson) Newsgroups: net.lang.prolog Subject: Re: eliminating duplicate solutions in Prolog Message-ID: <1164@sicsten.UUCP> Date: Sat, 31-May-86 01:52:16 EDT Article-I.D.: sicsten.1164 Posted: Sat May 31 01:52:16 1986 Date-Received: Tue, 3-Jun-86 04:42:50 EDT References: <126@sbcs.UUCP> Reply-To: lhe@sicsten.UUCP (Lars-Henrik Eriksson) Distribution: net Organization: Swedish Institute of Computer Science Lines: 31 In article <126@sbcs.UUCP> debray@sbcs.UUCP writes: >.... I'm thinking of the rather >mundane fact that any "real" system, to be useful, must interact with the >outside world, and hence necessarily have side effects like "read" >and "write". ..... I/o must do side effects, of course, but it is quite possible to hide this from a logic program, so that it appears to be in a completely "logical" environment. Example: Input could be done in a logical fashion by having a special kind of list with elements corresponding to successive objects read from the outside world. That is, the list would initially be an uninstantiated variable. Attempts to use the variable would cause an object to be read in and the variable would be bound to a pair of the first element and another uninstantiated variable. Attempts to use the rest of the list would cause the process to be repeated. That is, to the program, the list would look like it had everything read in on it from the start, but actually things would be read in only as they were needed. Of course, special precautions would have to be taken to make sure this worked when the program backtracked, but it is quite possible. (In fact, I have implemented input working in this way). Output could be done in a similar way, with objects being output as a list was bound. In both cases the program would think it was processing or creating a list, while it was actually reading or writing.