andrew <andrew(dot)ylzhou(at)gmail(dot)com> writes:
> I got errors in this query. I have a function complete(record) which
> takes a generic record type data. But it seems cannot be applied to a
> sub-select result:
As I seem to recall having mentioned several times already, PG 7.3 is
really, really weak in this area. (If you ever got it to work, which
I doubt you will, you'd probably then start to notice how badly it
leaks memory for whole-row operations...)
You should either forget this idea or invest the effort to move up
to PG 8.1.
regards, tom lane