Xref: utzoo comp.lang.misc:7073 comp.object:2886 Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!usc!wuarchive!hsdndev!husc6!bunny!harvey.gte.com!mto From: mto@harvey.gte.com (Tamer Ozsu) Newsgroups: comp.lang.misc,comp.object Subject: Re: CHALLENGE: heterogeneous collections Message-ID: <10839@bunny.GTE.COM> Date: 26 Mar 91 14:11:21 GMT References: <1991Mar22.210725.29448@neon.S Sender: news@gte.com Followup-To: comp.lang.misc Organization: GTE Laboratories, Waltham Lines: 45 In article pallas@eng.sun.com (Joseph Pallas) writes: >The problem is, no one has ever come up with a convincing reason why I >should want my type system to handle the heterogeneous collection. >This is because no one has come up with a convincing reason why I >should want to write programs that contain heterogeneous collections. >Needing collections of otherwise-unrelated objects generally signals a >flaw in the design, not a failing of the type system. > Well, let me see how I can take a position on both sides of this argument. I must admit that I missed the earlier segments, so if I am reiterating something that has been already addressed, my apologies. There are cases in object-oriented database systems where the results of a query is a heterogeneous collection. Object algebras take as input sets of objects and produce as output sets of objects (sometimes new objects that were not in the database to begin with -- e.g., explicit join -- but that is not important here). It is quite possible that the result set of objects is a heterogeneous collection. So, there is a need for the ability to handle heterogeneous sets of objects. On the other hand, I don't see why a type system and type checking mechanism cannot be developed to handle these collections. Some object algebras impose what is called union compatibility on the algebra operators [Shaw and Zdonik paper in the 6th Data Engineering Conference, 1990]. Union compatibility requires that members of the sets being operated on to be instances of types which are in a subtype relationship with one another. However, this may be too restricted. In last year's OOPSLA (1990), we presented a paper [pages 224-233] that discusses a more general type system for a simplified object algebra. The paper also discusses the issues of heterogeneous sets. This is by no means the definitive word on the subject, but it demonstrates that a type system, which is sometimes more restricted than one may like, can be developed. ==Tamer -- M. Tamer Ozsu Telephone: (617) 466-2098 GTE Laboratories Fax: (617) 290-0628 40 Sylvan Road Internet: mto@gte.com Waltham, MA 02254