Re: Efficiency; advanced/future SQL constructs

From: David Cressey <david_at_dcressey.com>
Date: Wed, 22 Aug 2001 13:50:13 GMT
Message-ID: <pMOg7.289$Iw2.19171_at_petpeeve.ziplink.net>


> Given the complexities in the syntax for CREATE, ALTER and DROP, I
> don't really consider them three statements. I consider CREATE TABLE a
> different statement than CREATE INDEX, for instance; otherwise, I note
> that we could get rid of CREATE and DROP by making them subclauses of
> a single ALTER DATABASE statement.

Point taken. And, IIRC, Codd was never a great fan of SQL to begin with. Of all the pieces of the relational puzzle, SQL was the one area where adhocracy prevailed of careful foresight. And SQL is the only part that many practitioners ever see! They even call these things "SQL databases".

If we get into redesigning the language, I'm sure lots of things would come up.

But I'll stand on my comment that a strong case would have to be made to add another construct to the language.

For an example of language add ons, consider Oracle's CONNECT BY, and Joe Celko's reaction to that feature.
CONNECT BY is so useful to me that I make use of it, Joe's objections
notwithstanding. But the question of whether CONNECT BY (or some equivalent) should be added to the SQL standard is a whole separate matter.

--
Regards,
    David Cressey
    www.dcressey.com
Received on Wed Aug 22 2001 - 15:50:13 CEST

Original text of this message