Sorry to resurrect an old thread, but I was at the Very Large Databases 
conference in Toronto earlier this week, and some of these very issues 
came up:

On Aug 25, 2004, at 5:23 PM, Ken North wrote:

>
> . In reality, the core technology of
> several "post-relational" products was developed years before the
> seminal papers by David Childs (1968) and E.F. Codd (1969).

   Alon Halevy's keynote Wednesday asserted that XML is causing as many 
problems for data integration as it solves,  mainly because the 
industry got so far  ahead of research.  He said that the challenge of 
designing an "internal algebra" for XML as one of the key issues that 
need to be resolved to get XML, relational, and textual data integrated 
more cleanly.  David Childs is also at the conference, and perhaps 
intriguingly, that is exactly what he is trying to do these days ... 
see http://xsp.xegesis.org .  The "extended set processing" stuff, in 
his view, can be seen as  an "internal algebra" for the pure relational 
model, SQL DBMS as they actually exist, and XML.


> Cache, for example, is an extended implementation of M (ANSI 
> X11.1-1995,
> withdrawn in 2002).  M is the '90s version of MUMPS, which was 
> developed at Mass
> General by Dr. Robert Greenes and company. It first ran in 1966 and was
> presented at the Fall Joint Computer Conference in 1969. MUMPS was 
> published as
> a standard by NBS in 1975 and ANSI in 1977, years before the first SQL 
> standard
> (1986).
>

For what it's worth, I asked the people at the Cache table at VLDB 
about this, and they deny it :-)  The company evolved from being a 
MUMPS vendor, but the Cache product was written from the ground up as 
an OODBMS, although of course that term is marketing poison these days.

That being as it may, I fully agree that "post relational" technology 
largely evolved from pre-relational technology such as MUMPS, Pick, 
Adabas, and probably all sorts of other things.