At 10:07 PM 5/31/01 -0400, David Jencks wrote:

>
> Xml provides the ability to easily construct hierarchical structures, which
> However, they have the disadvantage that
> queries usually require knowing something about the access path to the data
> Limitations with this model were perhaps the main
> So if you
> are planning to put data you need to query in the xml, you have two levels
> of problems:
> 1. You can't use relational query operations because the data is in a blob
> 2. You can't use relational query operations because the data is in an xml
> structure.
>

It is very interesting to see ideas come and go then return again.

The earliest database management system I know of was of what might
be called the amorphous data model -- a database was a collection
of records, each consisting of totally arbitrary attribute/value
pairs. It had some nice retrieval characteristics, but systematic
processing wasn't in it. Then came the hierarchical systems,
IBM's IMS and the ARPA Datacomputer. The hierarchical systems
were regular and the tree architecture seemed intuitive, but
updates were a nightmare and complex structures almost impossible
to model. Then came the CODASYL "network" data model of records
and named "sets". It could model anything, was readily updatable,
but was utterly impossible to layer a query language on. Then
came Codd's relational database (for two points, can anyone
explain the name?) which was highly regular, easily updated,
expressive, extensive, and easily retrievable. So good,
in fact, that once the performance problems were worked out
it put absolutely everything out of business.

Then the object data model showed up -- inheritance, polymorphism,
so very OO. The venture guys threw great big sacks of money
at them. But scrape off the hype and you find a warmed over
CODASYL database with all of the warts, glitches, and gotchas
that resulted in a mass extinction a decade earlier. Another
mass extinction.

So now, whoppee, another hierarchical representation, complete
with all the problems that made IMS such a mega-pig. XML as
an operational data representation is not just a terrible idea,
it's the reincarnation of a very famous terrible idea.

I'm now waiting for organic medium database management system.
Based by an all natural renewable resource (farmed trees
fueled by our sun) using an offshoot of optical storage
technology -- hole/no hole -- arranged in a user friendly
intuitive format of 12 rows and 80 columns.

When will they ever learn?
When will they ever learn?

Jim Starkey