Re: Regarding Plan tree output(Index/Bitmap Scan) - Mailing list pgsql-hackers

From Justin Pryzby
Subject Re: Regarding Plan tree output(Index/Bitmap Scan)
Date
Msg-id ZDYKY7IOESpiT2bW@telsasoft.com
Whole thread Raw
In response to Regarding Plan tree output(Index/Bitmap Scan)  (Ajay P S <ajayps547@gmail.com>)
List pgsql-hackers
On Tue, Apr 11, 2023 at 06:09:41PM -0700, Ajay P S wrote:
> I am trying to understand the Plan tree for select queries. Can you
> please help me with the below queries?
> 
> 1) Why is there a difference in plan tree for these two queries? User
> table tidx1 has an index on column 'a' .

Based on the query planner's cost estimate of the different scans.

> 2) Why do we do Index scan and not  Bitmap Index Scan for catalog tables?

There's no reason why it can't happen in general.

But you queried pg_class on a unique column, returning at most one row.
A bitmap couldn't help by making the I/O more sequential.  It can only
add overhead.

You can compare the costs of various plans by running EXPLAIN with
various enable_* GUCs to off.

BTW, your question should be directed to another list - this list is for
bug reports and development.

-- 
Justin



pgsql-hackers by date:

Previous
From: Stephen Frost
Date:
Subject: Re: longfin missing gssapi_ext.h
Next
From: Stephen Frost
Date:
Subject: Re: longfin missing gssapi_ext.h