Re: Two meanings of "data structures"
Date: Tue, 16 Nov 2004 23:23:29 -0500
Message-ID: <1b0r62-hco.ln1_at_pluto.downsfam.net>
Marshall Spight wrote:
> "Kenneth Downs" <firstinit.lastname_at_lastnameplusfam.net> wrote in message
> news:8oik62-3vd.ln1_at_pluto.downsfam.net...
>> I wonder if the literature has any well-known or popularly accepted works >> that map the data structures that we learn about in coding classes -- >> those being arrays, linked lists, hash tables and so forth -- to the >> relational model. >> >> By this I do not mean books that explain techniques in one or the other, >> there are plenty of those, but rather something that provides a mapping >> between them. More specifically, the modelling of lists and arrays and >> such in RDM.
>
> This is not an answer to your question, but I've wondered about
> a similar question. In Java, for example, one models everything with
> objects, and one has pre-built objects that model various generic
> data structures. If one wants a linked list, one uses
> java.util.LinkedList, etc. etc. In contrast, with RDM, one has to know how
> to model, for example, a tree with an adjacency list, or with a
> materialized path, etc.
If one wants a linked list in Java, one uses Perl :)
But seriously, I've been compiling my own maps of these, such as when you need an array, when a list, etc. The danger with this, and the reason for my post, is the fear of developing a lot of my own terms when others are in general use abroad.
I think Perl's hippie-free-love attitude does reveal how few structures there really are. You either point to something with an index or a name, and any thing can be an atomic value or another bag of things. Order is there but you can ignore it if is irrelevant, or change it.
SQL lacks a simple way to turn its native son, the table, into a list, something like the Join() function. It likewise lacks the Split(). Most procedural environments except for the may-as-well-be-dead Visual Foxpro lack any way to deal with tables intelligently.
>
> What I wonder about is, is there a way to achieve the drop-in property
> that OO gives you with RDM? It doesn't seem obvious how one
> would do that.
>
>
> Marshall
-- Kenneth Downs java.lang.String.tcpip.usenet.content.posting.sigblock.setSig("After finally finding the right object and method to set the sig block, I forgot what I wanted to say!");Received on Wed Nov 17 2004 - 05:23:29 CET