OT MySQL and PostgreSQL

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 02/01/07 22:19, Paul E Condon wrote:
> The recent heated discussion as to the relative merits of MySQL and
> PostgresSQL reminded me of a question that I want to ask of RDBMS
> experts, particularly experts who are willing to take a clear
> position. Namely, what do you think of the work of C. J. Date? He
> rejects SQL, as far as I can tell. Is there any support for this
> position in the real world? Or in the academic world? What of
> his objection to null 'values'? How does this play out when doing
> mission critical DB? Does it matter? Or are there standards
> techniques for avoiding any need for nulls?
>
> I look forward to reading an interesting discussion.

"SQL" isn't a complete/correct "expression" of relational algebra.
So, it can't express the full power of Codd/Date's theories.

As to support for this position in the real world, there's not much,
if any; SQL is Good Enough For Most Uses. As for nulls, I don't see
a problem with them. In fact, I think you need them, in order to
describe "unknown" data. Non-DBMS systems make you use a "special
value", but if any bit of data actually has that "special value",
you're hosed.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFFwtliS9HxQb37XmcRApU4AKDbLEZLPw7Q3RJGqn3tykMRNv5n4gCgpvBR
XBFBpSlYyngL9YT1Q6hp1RY=
=lewk
-----END PGP SIGNATURE-----

--

On Thu, Feb 01, 2007 at 09:19:48PM -0700, Paul E Condon wrote:
> The recent heated discussion as to the relative merits of MySQL and
> PostgresSQL reminded me of a question that I want to ask of RDBMS
> experts, particularly experts who are willing to take a clear
> position. Namely, what do you think of the work of C. J. Date? He
> rejects SQL, as far as I can tell. Is there any support for this

There doesn't seem to be a standard which everyone follows.

> position in the real world? Or in the academic world? What of
> his objection to null 'values'? How does this play out when doing
> mission critical DB? Does it matter? Or are there standards
> techniques for avoiding any need for nulls?

You can always use a default value constraint. I have found strange
results from my queries if using null values.

--
Chris.
======
Don't forget to check that your /etc/apt/sources.lst entries point to
etch and not testing, otherwise you may end up with a broken system once
etch goes stable.

--