Matthew <[EMAIL PROTECTED]> writes:
>> The correct fix is CommandCounterIncrement() in the DROP USER loop,
>> so that later iterations can see the changes made by prior iterations.
> Since postgre now suppport referential integrity and cascading deletes,
> wouldn't it make more sense to use that code to manage the relationship
> between pg_user and pg_group (and probably a wealth of other system tables),
Dunno if it's worth the trouble to change. Certainly this particular
bug would still be a bug, whether the cascaded deletion is done "by hand"
or not.
regards, tom lane
