On Mon, Nov 01, 2004 at 11:43:29PM -0600, Ed L. wrote:
> Andrew is going to need to say a little more before it's clear to me he's at 
> all justified in calling sequence triggers a terrible idea.  He mistated 
> the facts; sequences are NOT transactional (i.e., they're not rolled back).  
> Maybe that's just a typo on his part, I don't know.  He seems to presume 

Indeed it was a typo, as the context should make clear.  But since
it's apparently not obvious what my objection is, let me try to make
it so: if you have a trigger, it is performing an action.  Now, we've
already established that you don't want to grovel through a lot of
data every time you replicate, or this polling issue wouldn't even be
under discussion; and a sequence which was under heavy use would
cause a lot of notification events.  So, in any such system, you're
going to have to update the notification events in some way.  That
entails locking, because you can't have more than one process
flipping the same bit at the same time.  So you have a concurrency
problem.

The basic objection I have is that this moves the cost of keeping
track of the sequence from the asynchronous part, when we're
replicating, into the main transaction line, when data is being
transformed.  Given my experience with erserver and now Slony, I can
say that every single additional cost you impose in the main
transaction line will bite you later.  This is why Jan spent so much
time optimising the trigger for Slony -- the goal was to impose as
little overhead as possible.  To me, even a modest possibility of
contention on a sequence is a reason not to introduce such
functionality.  They're designed to be fast, and I don't want to lose
that.

> wants to do it, etc).  But fundamentally, async notification is very likely 
> the answer to the polling inefficiencies in one form or another.

I don't think anyone is disputing this; but a workable proposal is
what's wanted, I think.  I'd be delighted to be proven wrong in
what's above.

A

-- 
Andrew Sullivan  | ajs at crankycanuck.ca
This work was visionary and imaginative, and goes to show that visionary
and imaginative work need not end up well. 
		--Dennis Ritchie