Re: [HACKERS] update_pg_pwd

On Mon, 13 Dec 1999, Jan Wieck wrote:

>     Trigger  functions  should  allways  return  at  least a NULL
>     pointer of type HeapTuple, not be declared void. From this  I
>     assume it's an AFTER ROW trigger,

Must be after row, because it has to wait until the change is actually
written to pg_shadow. Better would be an AFTER STATEMENT is assume.

>     There  are already some exceptions coded into the test. These
>     are PL handlers. Since their real return value is  HeapTuple,
>     you  would  have  to  make  this  defined  special  type  not
>     selectable in another way. So why do you want?

I'm not sure I'm following you, but why would a function that doesn't have
a useful return value return one?

-- 
Peter Eisentraut                  Sernanders vaeg 10:115
peter_e@gmx.net                   75262 Uppsala
http://yi.org/peter-e/            Sweden