Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!haven!mimsy!alberto From: alberto@tove.umd.edu (Jose Alberto Fernandez R) Newsgroups: comp.lang.prolog Subject: Typing logic programs (WAS: RE: logic programs -> proced. lang) Message-ID:Date: 25 Sep 89 15:23:39 GMT Sender: nobody@mimsy.UUCP Distribution: comp Organization: University of Maryland Computer Sc. Lines: 27 In article <1503@irisa.irisa.fr> picart@irisa.irisa.fr (Marc Picart) writes: My opinion is that the compiler can infere a lot of the declarations by itself. (you can see the work of C. S. Mellish or S. K. Debray) I don't think that, in logic programming, it is necessary for the programmer to help the compiler. The reason is that the aim with high level languages, such as Prolog, is to separate the logic specification, which is the responsability of the programmer, from the implementation which is the responsability of the compiler. Well, you don't think that the domains of the objects that you are using is part of the specification? One of the usages that type checking has is to check if actually the usage of the objects correspond to the types that you assigned to them. Many programming errors can be catch by a compiler, when a proper use of types is done. Jose Alberto. -- :/ \ Jose Alberto Fernandez R | INTERNET: alberto@tove.umd.edu :| o o | Dept. of Computer Sc. | :| ^ | Univesity of Marylad | :\ \_/ / College Park, MD 20742 |