Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!att!tut.cis.ohio-state.edu!ucbvax!pasteur!hermes.Berkeley.EDU!aoki
From: aoki@hermes.Berkeley.EDU (Paul M. Aoki)
Newsgroups: comp.databases
Subject: Support for variable-length arrays of records
Message-ID: <9177@pasteur.Berkeley.EDU>
Date: 24 Nov 90 12:31:28 GMT
Sender: news@pasteur.Berkeley.EDU
Reply-To: aoki@hermes.Berkeley.EDU (Paul M. Aoki)
Distribution: na
Organization: University of California at Berkeley
Lines: 17

I've poked around a bit but have been unable to find much information
on DBMS support for variable-length record arrays (sequences of
records).  I'm not talking about variable-length attributes or
attribute arrays (like char[]); I'm interested in things like the
arrays supported by the EXTRA data model for EXODUS -- variable-length
arrays of first-class objects that can have new elements inserted into
arbitrary positions.  These are somewhat like LISP lists, really.
However, a DBMS that supported persistent LISP lists like
non-persistent LISP does, i.e., with cons cells, would really eat it
on lists of any length.

I imagine there are plenty of OODBMS folks who have dealt with this
issue.  What general approaches were used?  Or did *everyone* punt and
do something like stick all of the array elements (or object
pointers) inside another object?
--
    Paul M. Aoki   |   aoki@postgres.Berkeley.EDU   |   ...!ucbvax!aoki