Going back through the patches we had to make to 9.0 to move to
PostgreSQL triggers, I noticed that I let the issues raised as bug
#6123 lie untouched during the 9.2 development cycle. In my view,
the best suggestion for a solution was proposed by Florian here:
http://archives.postgresql.org/pgsql-hackers/2011-08/msg00388.php
As pointed out in a brief exchange after that post, there would be
ways to do the more exotic things that might be desired, while
preventing apparently straightforward code from doing surprising
things. I'm not sure whether that discussion fully satisfies the
concerns raised by Robert, though.
Because I let this lapse, it only seems feasible to go forward with
a patch for 9.2 if there is consensus around Florian's proposal. If
there is any dissent, I guess the thing to do is for me to gather
the issues and see about getting something into 9.3, once 9.2 work
has died down -- in five months or so. Wisconsin Courts can
continue to deal with the issues using my more simple-minded patch,
but others still are getting surprised by it -- bug #6226 is
apparently another manifestation.
Comments?
-Kevin