Re: Forcing GiST index to be used

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Dave Blasby <dblasby(at)refractions(dot)net>
Cc: pgsql-hackers(at)postgresql(dot)org
Subject: Re: Forcing GiST index to be used
Date: 2001-08-14 20:01:00
Message-ID: 19963.997819260@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Dave Blasby <dblasby(at)refractions(dot)net> writes:
> I have a custom datatype (the PostGIS geometry type), which I have
> indexed using a GiST index.
> The problem is, its difficult to get PostgreSQL to actually use the GiST
> index. The only way I can get it to be used is by 'set enable_seqscan =
> off', which seems a bit cheezy. What am I missing? Do I have to make
> some sort of amcostestimate() function or something?

What sort of selectivity estimator (oprrest entry) do you have attached
to the indexable operator? If there's no estimator, the default
selectivity is something like 0.5 --- way too high to cause an index to
be used.

regards, tom lane

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Steve Howe 2001-08-14 20:25:43 Retriving users from group ?...
Previous Message Tom Lane 2001-08-14 19:54:42 Re: int8 sequences --- small implementation problem