Actually, in fact there are such things (and why was this 
marked as a duplicate?)
[download]

While oracle is in fact at it's core a relational database structure, 
oracle8i and above provide some nifty objectish approaches
to both storing and recovering data in more complex structures.

Think of the fields represented here with *_NT as being tables
within the row elements of the interface table.  In perl you
would call this strucrure a multi dimensional hash, where the keys
*_NT point to embedded hashes as their values.

The upshot of doing something like this is that when you are
dealing with many tables which have extremely large counts of
row data, you do not have to do explicit joins to find the
data you wish to retrieve, thus making things run ALOT faster 
by avoiding the need to create a large cartesian product and
then eliminating rows before retrieving your data.

I cant possibly have been the first person to try this. <:)

   --perlboy
[download]

Okay, I read the blurb, and it sounds pretty grotesque.
Personally I'd just do two SQLs rather than gild a lily...
and not knit the code so closely to one database manufacturer.


I cant possibly have been the first person to try this. <:)
You aren't.. However, you'll not want to hear that..
From what I've read they're akin to LOBS, I guess you've checked this on CPAN?

--

Brother Frankus.

¤