Lambda the Ultimate

inactiveTopic SQL Limitations
started 6/13/2002; 9:06:08 AM - last post 6/17/2002; 9:40:56 AM
Ehud Lamm - SQL Limitations  blueArrow
6/13/2002; 9:06:08 AM (reads: 2023, responses: 4)
SQL Limitations
Patrick Logan explains why he doesn't like SQL.

As you all know I don't really like language comparisons (I am not the only one), and usually find discussions of language limitations rather boring. But Patrick raises some interesting issues that may interest others.

By the way, I don't think the reason for SQL's success is that it was desgined by IBM. SQL (with some additions) survived much better than PL/I, for example, which was designed and promoted by IBM. But this really is not the most important part of Patrick's analysis

Posted to Logic/Declerative by Ehud Lamm on 6/13/02; 9:10:00 AM

Patrick Logan - Re: SQL Limitations  blueArrow
6/13/2002; 9:20:39 AM (reads: 1249, responses: 1)
To clarify my opinion on SQL's success... I tried to say that its success is due not just to IBM's endorsement, but also the fact that Oracle picked up the ball from there. Oracle believed the SQL/relational market would be huge because IBM was behind it.

So Oracle applied all their start-up energy toward that end. And I believe because of Oracle's ability as one of the most successful software start-ups of all time, they essentially turned SQL into the success they thought IBM would itself be establishing.

Chris Rathman - Re: SQL Limitations  blueArrow
6/13/2002; 9:56:17 AM (reads: 1235, responses: 0)
I'm working on two database projects at the moment. One of them relies heavily on time sequencing. The value of a column is dependent on the day you are querying with the need to look backwards and forewards. In addition, historical data may need to be corrected in the future and the difference between the previous value and the current value for that date has to be reconciled. The application is using a time series or sorts, though the periodicity is anything but regular.

The other application I do is a tracking aplication that tries to keep tabs on how effectively an organization is meeting their commitments. In order to model an organization, you have to use a hierarchical framework (a tree). Also, the access rights for the tree have to be assigned in a tree themselves (directory rights).

Anyhow, I agree that SQL is severely handicapped when it comes to time series and tree data. I spend an inordinate amount of my time getting around these very issues. They are not intractable if you have stored procedures and views, but they are pain none-the-less.

Ehud Lamm - Re: SQL Limitations  blueArrow
6/13/2002; 11:55:58 AM (reads: 1271, responses: 0)
I didn't research this, but from what I remember SQL was the thing to know if you worked databases, before Oracle became high profile. That's personal experience, not research.

Perhaps it is true that Oracle's support for SQL ensured its continued prominence. But it can be the other way around as well..

Leandro - Re: SQL Limitations  blueArrow
6/17/2002; 9:40:56 AM (reads: 1141, responses: 0)
This is not about language comparisons, but fundamental mistakes in SQL -- many of them, and then some more, are actually also present in most OO languages, and OO RDBMSs usually merge the worse of both.

Until data sublanguages based on the relational model become popular, we're stuck with many such pains. As far as I know there's only one site,, really dealing with this ugly situation.