On Fri, 12 Dec 2003, cnliou wrote:
> It looks to me a referential integrity problem (only?)
> within PLPGSQL. Plesase see the test result below.
There have been discussions in the past about when cascade events
should occur. The code currently does what I believe was last
agreed upon, although its behavior is fairly wierd for deferred
constraints and functions. Right now the cascade happens at the
end of the full statement (in this case the call to the function)
which is why you get a key constraint error in the second call to f1
and why the later inserted row is removed in f2.