Re: First Impressions on Using Alphora's Dataphor

From: Christopher Browne <cbbrowne_at_acm.org>
Date: 27 Aug 2004 02:39:56 GMT
Message-ID: <2p7l7sFhrhjuU1_at_uni-berlin.de>


In the last exciting episode, lajos.nagy_at_gmail.com (Josh Hewitt) wrote:
> Sorry for the long post, but I felt that this might be an
> interesting topic to discuss in this news group.

No, that's a very interesting set of observations.

The challenges involved in the transformations of domains as you experiment with them is interesting indeed.

I'm seeing, in this, a tad of need for some dynamic of type transformations. Two things leap to mind as opposite ends of the spectrum:

  • The CLOS notion of defined "classes," which actually provide a way to transform instances into the new form should you happen to redefine them.

   <http://www.ai.mit.edu/projects/iiip/doc/CommonLISP/HyperSpec/Body/stagenfun_upd_efined-class.html>

   That's object-oriented, not relational, but it sure points to a way    to update "tuples" on demand...

  • The ML notion of polymorphism, where types can be left abstract somewhat until the last moment; you can define functions that are strongly typed, but which do not care what they look like 'til the moment that you actually call the function.

I'm not sure either is the ultimate answer, and it would _definitely_ be an inconvenience to make a little domain change "over here" that leads to 200GB worth of tuples having to be revised "over there."

-- 
select 'cbbrowne' || '_at_' || 'acm.org';
http://www.ntlug.org/~cbbrowne/spreadsheets.html
"We defeated the enemy with teamwork and the hammer of not bickering."
-- The Shoveller, Mystery Men
Received on Fri Aug 27 2004 - 04:39:56 CEST

Original text of this message