RE: First draft of PG 19 release notes

Phil Florent <philflorent@hotmail.com>

From: Phil Florent <philflorent@hotmail.com>

To: Andreas Karlsson <andreas@proxel.se>, Bruce Momjian <bruce@momjian.us>, PostgreSQL-development <pgsql-hackers@lists.postgresql.org>

Date: 2026-05-01T13:24:51Z

Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

doc PG 19 relnotes: more fixes
- 8656ba7f7187 19 (unreleased) landed
doc PG 19 relnotes: various corrections
- 1d751b4b6b01 19 (unreleased) landed
doc PG 19 relnotes: adjust item to mention pg_replication_slots
- cfedd4513357 19 (unreleased) landed
doc PG 19 relnotes: remove "Add fake LSN support to hash index"
- 41b60bf172ef 19 (unreleased) landed
doc PG 19 relnotes: add two optimizer hooks
- cac0f24eb57f 19 (unreleased) landed
doc PG 19 relnotes: remove "Optionally" for CPU optimizations
- 06fccab4c61b 19 (unreleased) landed
doc PG 19 relnotes: adjustments/removal of items
- 8974a7c433d3 19 (unreleased) landed
Use fake LSNs to improve nbtree dropPin behavior.
- 8a879119a1d1 19 (unreleased) cited
Clean up ICU includes.
- af2d4ca191a4 19 (unreleased) cited

Hi,
I see on many UTF-8 customer bases expression-based index on lower(col) to fulfill some ORM implementation (lower(col) = lower(exp) or "ilike" but ilike is even more "difficult" to index so lower(col) is not so bad...) If pg 19 makes lower(utf-8_text) faster, it's definitely something I would be happy to read in a major upgrade .
Best regards,
Phil

________________________________
De : Andreas Karlsson <andreas@proxel.se>
Envoyé : jeudi 30 avril 2026 09:16
À : Bruce Momjian <bruce@momjian.us>; PostgreSQL-development <pgsql-hackers@lists.postgresql.org>
Objet : Re: First draft of PG 19 release notes

On 4/30/26 9:08 AM, Andreas Karlsson wrote:
> I noticed that two of my performance patches were not included in the
> release notes and I personally think one of them belongs in the release
> notes while other not.
>
> The one I think belongs is the one below since it it really results in
> major speedups of lower(), upper(), initcap() and casefold() on ICU with
> UTF-8. Not having to convert from UTF-8 to UTF-32 and back is often a
> major speedup.

Saw the big discussion on performance improvements if they should be
included or not. So I will just leave this alone. :)

--
Andreas Karlsson
Percona