Ann W. Harrison wrote:

>>
> Here's a SQL standard version of the same problem:
>
> SQL> select count (*) from t;
>
> COUNT
> ============
>
> 50
>
> SQL> delete from t t1 where t1.id in
> CON> (select min(t2.id) from t t2);
> SQL> select count (*) from t;
>
> COUNT
> ============
>
> 0
>

Just in case you think that the results of that delete are correct, I've
taken the same database, inverted the order so the highest values appear
first in storage order and re-run the query.

SQL> rollback
CON> ;
SQL> update t set id = 60 - id;
SQL> select count (*) from t;

COUNT
============

50

SQL> delete from t t1 where t1.id in
CON> (select min(t2.id) from t t2);
SQL> select count (*) from t;

COUNT
============

49


Depending on storage order, the same query deletes one row or 50. Not
terribly good. And InterBase does the same thing. And they don't have
a solution because they've made verb undo lists optional.

Regards,


Ann