Re: Array of Integer

aland@informix.com (Colonel Panic) writes: >In article <1n74h1INNst7@access.digex.com> grimes@access.digex.com (Seth Grimes) writes: >>InterBase (from Borland) and Progress are both RDBMSes with an array >>datatype. I'd be interested in seeing more of an explanation of why this >>violates normalization. I (also) have the view that an array can be >>viewed as a single entity akin to a scalar field. >The first rule of normalization is to get rid of repeating groups (e.g. >arrays). This is the basic requirement to get to 1NF. Would you please explain why you think an array of numbers is any more a repeating group than a text string? To answer my own question, an array usually stores a set of function values for some independent index value. For instance, a time series is a set of measurements as a function of time: y[i] = f(t[i]). But on the other hand, a text string is (especially if you come from a C-type point of view) really an array of character*1 and is essentially equivalent to a time series if you transform your thinking from "index domain" to "time domain." How about it? Perhaps the real answer is that I'm not a theorist or a vendor, interested in either defending my academic standing or in showing how the "extended" but useful features of competing products violate the Twelve Commandments. I've learned that, in production settings, one often has to violate normalization for performance or based on application logic. (As a systems analyst, I try to sell management on the idea that first one looks at business requirements, then at software that will answer the requirements, and then at hardware.) If an array datatype fits easily into data management requirements, who cares if there's a debate going on whether it constitutes a repeating group and/or violates 1NF? Seth * * * * * * * * * * * * * * * * * * * * * * * * * * * * Seth Grimes, Consultant Directorate for Computers and Communications Organisation for Economic Co-operation and Development OCDE-DCC/SDD phone: (33-1)45.24.19.48 2, rue Andre-Pascal fax: (33-1)45.24.19.51 75775 Paris, France Internet: grimes@oecd.fr