Re: Why are data types size limited?

joe_celko_at_my-deja.com wrote:
>
> >> From an ignorant-user standpoint, they {SQL databases} are great
> except for field size limitations ... Why is this? Why can't databases
> be written to handle arbitrarily long strings, and grow or shrink all
> input fields as needed? <<
>
> That is actually a good question. When we were doing the SQL-86 and
> SQL-89 standards, we knew that SQL really stands for "Scarely Qualifies
> as a Language" -- a little ANSI humor.
>
> The specs do not talk about internal representation of data types in
> the SQL database, only scale and precision, character sets, and other
> abstract things. The reason is that you never directly see SQL data!!
> The database exists in one place, the host program exists in another
> place and the SQL engine passes data back and forth between them,
> converting datatypes on the fly. This is why 'C' and Cobol can use the
> same database.
>
> The datatypes we picked were the most universal, easiest to map to host
> programming languages we could imagine.
>
> --CELKO--
>
> Sent via Deja.com http://www.deja.com/
> Before you buy.

That reminds me that the original definition of SQL, back in the lab at Sta.Teresa (IBM) was Simplified (or Simple-minded ;-) Query Language. It was intended for management types who couldn't think logically to save their souls (if they have one). Of course, it quickly became so obtuse that no reasonable person can express any reasonably complex relationship with it. I often bemoan the fact that QUEL wasn't more widely adopted. It may not have been particularly great, but it was much better than SQL. However, the one query language which I really liked was ZIM, which supports full ER models. Really powerful set processing, and implied join (including recursive and self join) semantics via the use of defined relationships.

Bill Boyle
woboyle_at_ieee.org Received on Wed Apr 12 2000 - 00:00:00 CEST