On Mon, 29 Jul 2002, Roderick A. Anderson wrote:
> I would like to add one other thought. There are many web site
> designers that get thrust into being a web site programmer. Without
> an understanding of database design and a novice programmers (?) view
> of the process the benefits of letting the database (RDBMS) do the
> database work isn't recognized. They code it all in the CGI.
Well, I'll add two points to this, then:
1. Often there's a lot more benefit to moving the work from the database
to the application structure. Database schemas are hard to change, and
hard to keep under revision control. When I was doing a large website,
it was much, much easier to say "everything goes through these Java
classes" than "everything goes through the database." I could change the
database schema at will and know that my data was safe, because I could
have old interfaces running simultaneously with new.
(Though I'll admit, good view support would have mitigated this problem
quite a lot. But there is *no* database in the world that has really
good view support; they all fail on various updates where one can
theoretically do the Right Thing, but in practice it's very difficult.
And I don't think that's going to change any time soon.)
2. I expect that even most PostgreSQL--or even database--experts don't
have a real understanding of relational theory, anyway. That we still
have table inheritance shows that. As far as I can tell, there is
nothing whatsoever that table inheritance does that the relational model
does not handle; the whole "OO" thing is just another, redundant way of
doing what we already ought to be able to do within the relational model.
I'm still waiting to find out just what advantage table inheritance
offers. I've asked a couple of times here, and nobody has even started
to come up with anything.
All that said, though, don't take this as any kind of a dismissal of
postgres. It's in most ways better than MySQL and also some commericial
systems, and many of its failures are being addressed. Postgres for some
reason seems to attract some really, really smart people to work on it.
If I could see something better, I'd be there. But I don't.
cjs
--
Curt Sampson <cjs@cynic.net> +81 90 7737 2974 http://www.netbsd.org Don't you know, in this new Dark Age, we're
alllight. --XTC