Re: matrix encoding IS adjacency list

From: Marshall Spight <marshall.spight_at_gmail.com>
Date: 19 Sep 2005 21:28:04 -0700
Message-ID: <1127190484.053528.295470@f14g2000cwb.googlegroups.com>

Vadim Tropashko wrote:

> Which is equivalent to saying: "Let's represent a node with both
> materialized path and truncated path". Surely, there is a referential
> integrity constraint between these two columns. The point is that you
> artificially made that second column.

The materialized path approach and other approaches to handling trees in SQL are intellectually and academically interesting, but as practical techniques, they feel to me like they are working hard to compensate for shortcomings in the language. I suppose this is not a particularly original observation.

The thing is, I haven't got a clear idea of what the alternatives are. Tree handling in OOPLs works fine, I suppose; at least it's less cumbersome that SQL. Many functional programmers are quite happy with using pattern matching and algebraic data types. This is certainly much better, but still doesn't feel like it has hit bedrock yet.

Anyone have any candidates? Term rewriting languages? Rho calculus? Anything? I would love to have something that feels as solid for trees as the relational algebra does for relations.

Marshall Received on Mon Sep 19 2005 - 23:28:04 CDT