| From: | Aleksander Alekseev <aleksander(at)tigerdata(dot)com> |
|---|---|
| To: | PostgreSQL Development <pgsql-hackers(at)postgresql(dot)org> |
| Cc: | John Naylor <johncnaylorls(at)gmail(dot)com> |
| Subject: | [PATCH] Simplify SortSupport implementation for macaddr |
| Date: | 2026-02-25 13:05:46 |
| Message-ID: | CAJ7c6TMk10rF_LiMz6j9rRy1rqk-5s+wBPuBefLix4cY+-4s1w@mail.gmail.com |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Hi,
Since Datums are 64-bit values and MAC addresses have only 6 bytes,
the abbreviated key contains the entire MAC address and is
authoritative, as pointed out by John Naylor [1]
This fact eliminates the need for cardinality estimation using
HyperLogLog since macaddr_abbrev_convert() is dirt cheap and not
lossy. There are no reasons to give up on abbreviation.
Potentially we could go even further and pass MAC addresses by value
rather by reference [2]. This would eliminate the need of abbreviation
completely since SortSupport->comparator could just compare two
Datums, as we do for Timestamps. This is a more invasive change though
that deserves more discussion and thus not proposed here.
[1]: https://postgr.es/m/CANWCAZYWdOEnoL_88VpMge1RtRpBz-VRCjdcu-eA4q3U6LvpDw%40mail.gmail.com
[2]: https://postgre.es/m/CAJ7c6TM8up%3DYih8pRLPy4wHwLzHf7w22tQ80-8ZBm__E%3D8_5HA%40mail.gmail.com
--
Best regards,
Aleksander Alekseev
| Attachment | Content-Type | Size |
|---|---|---|
| v1-0001-Simplify-SortSupport-implementation-for-macaddr.patch | text/x-patch | 5.8 KB |
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Zsolt Parragi | 2026-02-25 13:09:21 | Re: centralize CPU feature detection |
| Previous Message | Marcos Pegoraro | 2026-02-25 12:59:38 | Re: Partial Mode in Aggregate Functions |