Ref Cursor Overhead

What type of overhead is involved in using ref cursors vs. implicit cursors? I have developers that insist on using ref cursors for output variables instead of individual output variables even when returning result sets that will always contain a single record. They argue that they don't want to concern themselves with data types or the number of columns being returned. They only want to deal with record sets.

Besides the fact that this type of programming seems sloppy to me (just an opinion), I have argued that there is additional overhead involved in using ref cursors and that they should be reserved for where they are truly needed like when returning multi-record result sets or when using native dynamic SQL.

Is this an argument worth pursuing? Is there enough additional overhead involved in utilizing ref cursors that I should enforce that they be used sparingly?

I have read in a few places that ref cursors require a good deal of overhead vs. implicit cursors, but I'd like to confirm the fact.

Any input that anyone could provide would be greatly appreciated.

Thanks. Received on Mon Jun 10 2002 - 10:47:13 CDT