--On fredag, januari 14, 2005 11.52.38 +1100 John Hansen
>> > Dunno if you're desperate enough to try that ... but it does seem to
>> > work.
> if yo're going to hack anyway, then why not just simply tell the planner
> that you know better and that it should use the index, regardles of the
> stats collected?
> set enable_seqscan=false;
> <your original query here>;
> set enable_seqscan=true;
yeah, maybe I'll do that, but I stumbled on at least one more complicated
query that lost performance from disabling seq_scans. it had a union with a
query that gains performance from disabling seq_scans... heh...
Thanks for your time, guys! I now have some options to move on. I would
love to see this working better in a future version of postgresql, thought
In response to
pgsql-hackers by date
|Next:||From: Michael Fuhr||Date: 2005-01-14 01:25:13|
|Subject: Re: Slow PL/pgSQL 8.0.RC5 (7.4.6. 3times faster)|
|Previous:||From: John Hansen||Date: 2005-01-14 00:52:38|
|Subject: Re: Bug? 8.0 does not use partial index|