Horton

Re: List of Bitmapset (was Re: ExecRTCheckPerms() and many prunable partitions)

Amit Langote <amitlangote09@gmail.com>

View thread View on postgr.es
From: Amit Langote <amitlangote09@gmail.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Richard Guo <guofenglinux@gmail.com>, Alvaro Herrera <alvherre@alvh.no-ip.org>, David Rowley <dgrowleyml@gmail.com>, Greg Stark <stark@mit.edu>, Julien Rouhaud <rjuju123@gmail.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2022-11-16T02:35:08Z
Lists: pgsql-hackers 
On Wed, Nov 16, 2022 at 3:25 AM Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Amit Langote <amitlangote09@gmail.com> writes:
> >  On Mon, Nov 14, 2022 at 11:57 PM Tom Lane <tgl@sss.pgh.pa.us> wrote:
> >> + * The new member is identified by the zero-based index of the List
> >> + * element it should go into, and the bit number to be set therein.
>
> > The comment sounds a bit ambiguous, especially the ", and the bit
> > number to be set therein." part.  If you meant to describe the
> > arguments, how about mentioning their names too, as in:
>
> Done that way in the patch I just posted.

Thanks.

> >> +   /* forboth will stop at the end of the shorter list, which is fine */
>
> > Isn't this comment unnecessary given that the while loop makes both
> > lists be the same length?
>
> No, the while loop ensures that a is at least as long as b.
> It could have started out longer, though.

Oops, I missed that case.

The latest version looks pretty good to me.

-- 
Thanks, Amit Langote
EDB: http://www.enterprisedb.com

Commits

  1. Invent "multibitmapsets", and use them to speed up antijoin detection.