Horton

Re: Making empty Bitmapsets always be NULL

David Rowley <dgrowleyml@gmail.com>

View thread View on postgr.es
From: David Rowley <dgrowleyml@gmail.com>
To: Yuya Watari <watari.yuya@gmail.com>
Cc: Tom Lane <tgl@sss.pgh.pa.us>, pgsql-hackers@lists.postgresql.org
Date: 2023-07-03T00:09:48Z
Lists: pgsql-hackers

Attachments

On Mon, 3 Jul 2023 at 09:27, David Rowley <dgrowleyml@gmail.com> wrote:
> If nobody else wants to take a look, then I plan to push the v4 + the
> asserts in the next day or so.

Here's the patch which includes those Asserts.  I also made some small
tweaks to a comment.

I understand that Tom thought that the Asserts were a step too far in
[1], but per the bugs found in [2], I think having them is worthwhile.

In the attached, I only added Asserts to the locations where the code
relies on there being no trailing zero words.  I didn't include them
in places like bms_copy() since nothing there would do the wrong thing
if there were trailing zero words.

David

[1] https://postgr.es/m/2686153.1677881312@sss.pgh.pa.us
[2] https://postgr.es/m/CAJ2pMkYcKHFBD_OMUSVyhYSQU0-j9T6NZ0pL6pwbZsUCohWc7Q@mail.gmail.com

Commits

  1. Remove trailing zero words from Bitmapsets

  2. Remove local optimizations of empty Bitmapsets into null pointers.

  3. Require empty Bitmapsets to be represented as NULL.

  4. Mop up some undue familiarity with the innards of Bitmapsets.

  5. Remove bms_first_member().