Re: Re: making write location work (was: Efficient transaction-controlled synchronous replication)

On Wed, Mar 23, 2011 at 8:20 PM, Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
> On Wed, Mar 23, 2011 at 3:33 PM, Simon Riggs <simon(at)2ndquadrant(dot)com> wrote:
>> In any case, that's not the only argument for keeping it. We introduce
>> the view in this release and I would like it to stay the same from
>> now, since we know we will need that info later.
>
> At least as I understand it, it's not our project policy to carry
> around code that doesn't accomplish anything useful.  I have no
> objection to keeping the field; I simply think that if we're going to
> have it, we should make it work, as in fact it did before you changed
> it without discussion.  You haven't offered any evidence at all that
> it introduces any kind of a performance regression AT ALL, much less
> that such any such regression can't be trivially patched around by
> making SyncRepReleaseWaiters exit quickly if the flush LSN hasn't
> advanced.  The onus is as much on you to justify the change as it is
> on me to justify changing it back.

What a stupid conversation.

There's no onus on me to have to keep justifying to you why the code
is the way it is, but I do.

If you want to make a change that I already know reduces performance,
you have to have a good reason. So far, you don't.

Stop fussing and wrap the release.

--
 Simon Riggs                   http://www.2ndQuadrant.com/
 PostgreSQL Development, 24x7 Support, Training & Services