Re: [SQL] key set delete query

Michael McCarthy <michael(at)tcsi(dot)com> writes:
> I notice to my dismay that key set query optimization only works for
> SELECT :( This could be a significant performance problem for my
> application, which would like to do a key set based delete.

> Any thoughts on the advisability of changing the criteria for
> transformation of key set based queries?

I have no intention of putting any effort into KSQO, other than
removing it as soon as the rest of the optimizer does a reasonable
job with OR queries ;-). You could try removing the check for a
SELECT and see what happens, though.

> I know that cnfify is being reworked for release 7; any comment on when
> 7 will be available?

Current plan is to go to beta mode on 1 Feb. Final release, hard to
say. Marc will probably say "1 March for sure", but I lean more towards
the "when it's ready" approach...

> How stable is the latest beta? About 10 days ago I tried building from the
> daily snapshot, but didn't get very far.

We seem to be in the usual pre-beta flurry of committing half-tested
code :-(. I think the underlying stability of the system has been
amazingly good throughout this development cycle, but there have been
a lot of minor build and regress test problems in the past couple weeks.
If you have the know-how to get past those, things pretty much work.

regards, tom lane