Re: [xml-dev] functional databases noodling ...

[Michael Sokolov <sokolov@ifactory.com>]

On 2/15/2012 3:23 PM, Michael Kay wrote:
> One of the reasons O2 was a great project was that it took input from 
> the functional database research rather than (as with other object 
> databases) merely trying to be a persistent object-oriented 
> programming language.
>
> Michael Kay
> Saxonica
I worked on one of those (ONTOS) for a few years - it was my first real 
job, and it always seemed that we were struggling to find viable 
applications for it (and customers!)  Looking back now, I think the big 
missing link was a query language: something really data-oriented seemed 
to be called for. Instead ONTOS provided persistent C++ collection data 
structures and assumed that would be sufficient, but it wasn't, and it 
was just too hard to optimize every conceivable kind of C++ access to 
remote storage.  Our most successful competitor, Object Design, had 
developed a memory-mapped storage that seemed to be the best thing going 
in that arena, but even still I don't think it really handled joins and 
indexing as well as RDBs had been doing for years.

-Mike