| From: | Ranier Vilela <ranier(dot)vf(at)gmail(dot)com> |
|---|---|
| To: | Pg Hackers <pgsql-hackers(at)postgresql(dot)org> |
| Cc: | David Rowley <dgrowleyml(at)gmail(dot)com>, "watari(dot)yuya(at)gmail(dot)com" <watari(dot)yuya(at)gmail(dot)com> |
| Subject: | Re: Making empty Bitmapsets always be NULL |
| Date: | 2023-06-21 12:16:01 |
| Message-ID: | CAEudQArGkwRNu1RwKh+VKHq461MYWvuPetMXuv4M5A1NT=Asng@mail.gmail.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Hi,
David Rowley wrote:
>I've adjusted the attached patch to do that.
I think that was room for more improvements.
1. bms_member_index Bitmapset can be const.
2. Only compute BITNUM when necessary.
3. Avoid enlargement when nwords is equal wordnum.
Can save cycles when in corner cases?
Just for convenience I made a new version of the patch,
If want to use it.
regards,
Ranier Vilela
| Attachment | Content-Type | Size |
|---|---|---|
| remove_zero_trailing_words_from_bitmapsets_v5.patch | application/octet-stream | 22.3 KB |
| From | Date | Subject | |
|---|---|---|---|
| Next Message | James Coleman | 2023-06-21 12:45:19 | Re: Use of additional index columns in rows filtering |
| Previous Message | Andrew Dunstan | 2023-06-21 11:35:15 | Re: pgindent vs. pgperltidy command-line arguments |