Let me note that...

 - The "grovelling around the configuration" approach took a couple of
   days work to simply think about what *appeared* to be all of the
   complexities, forget about starting to implement them.

 - I started work on OMIT COPY yesterday afternoon, and had something
   passing tests this afternoon, and I wasn't working on it full time.

The difference in complexity between the approaches is pretty
staggering ;-).

There is a *bit* of complexity in that you need to have slonik scripts
around to re-initialize the cluster.  I've got a 3/4-there script in
tools/slonikconfdump.sh; it is missing two things that I'd like to
have before treating this as finalized:

 a) We need something to pull information about triggers out of the
    former sl_trigger table, and generate ALTER TABLE statements to apply
    to the appropriate nodes.

 b) There needs to be a topological sort of the subscriptions so as to
    do them in the correct order.
-- 
select 'cbbrowne' || '@' || 'linuxfinances.info';
http://linuxfinances.info/info/emacs.html
Bad command. Bad, bad command! Sit! Stay! Staaay...