Re: How to create a DBMS from scratch?

On 20 Nov 2002 08:02:20 -0800, Anthony.Youngman_at_ECA-International.com (Anthony Youngman) wrote:

>Apart from the terminology, the Pick mindset and the XML mindset are
>very similar, almost to the point of a simple 1-to-1 topological
>equivalence.

This is one of XML's biggest failings (and although this XML=Pick analogy isn't true, it's close enough)

Oracle 8 & 9 have a powerful XML storage mechanism, but this is itself pointlessly limited. Put simply, storing XML with a known XML Schema is a no-brainer in almost any DB technology, but it's just not an interesting problem, nor hardly a useful problem to solve.

XML is easy, but most of the data we receive (rather than create) isn't this well behaved. XML is structured, text is unstructured, and the real-world is by and large semi-structured, somewhere in the middle. We can represent this structure (RDF) and we can describe this representation (DAML) or even explain it (DAML+OIL). What we don't get though is prior notice when we build our application of how the data it receives in the future wil lbe structured. Our consumer app has to "make it up as it goes along" and be enormously fluid, not only about what it stores, but what it also permits queries upon.

As it happens, I work on just this problem and have built some big and _efficient_ RDF triple stores on top of commercial SQL's. Received on Thu Nov 21 2002 - 02:49:34 CET