Tom Lane wrote:
> > I've actually got code (that no longer cleanly applies, but...) that uses
> > the single query version with NOT EXISTS (which could be easily changed to
> > either of the other forms) and was planning to put it together for a patch
> > when 7.5 devel started because I figured it wasn't precisely a bug and
> > wouldn't get accepted for 7.4.
>
> Well, Bruce has this on his open-items list, so I figure we have a green
> light to do something for 7.4 if we can work out what to do.
I put it on because I wasn't clear exactly what was happening in the
discussion. There also was discussion that we want to improve this now
because everyone will be using for upgrading to 7.4, and with a ~50% db
reload speed improvement, it is hard to ignore.
I am not against the idea of adding it to 7.4 if we can do it cleanly,
and in fact we are sort of waiting for more serious bug reports at this
time, so doing something else to improve the code isn't out of the
question if we can do it without stumbling --- seems dump/reload gets
full attention only during beta, which makes sense.
However, I think we have to be honest that this is a performance
_improvement_, not a fix. Yea, you can say it was a bug that we did it
the way we did in the past, but you have to look real hard to see it
that way. :-)
Let's have multiple people eyeball the patch and give it an OK and we
can add it for 7.4 if people want it. If you look really hard, you can
say it is a fix for a missing pg_upgrade!
-- Bruce Momjian | http://candle.pha.pa.us pgman@candle.pha.pa.us | (610)
359-1001+ If your life is a hard drive, | 13 Roberts Road + Christ can be your backup. | Newtown Square,
Pennsylvania19073
