Re: Extensible sync handler registration (register_sync_handler)

From: Greg Lamberson <greg(at)lamco(dot)io>
To: pgsql-hackers(at)lists(dot)postgresql(dot)org
Subject: Re: Extensible sync handler registration (register_sync_handler)
Date: 2026-07-30 21:46:11
Message-ID: 20260730214611.2772955.lamco@lamco.io
Views: Whole Thread | Raw Message | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

A status update ahead of the commitfest close.

The v2 patches still apply cleanly to current master, and CFBot is green
across all platforms as of 2026-07-29: SanityCheck, Linux Meson (64-bit
and 32-bit), Linux Autoconf, macOS Meson, Windows Visual Studio, Windows
MinGW Meson, and CompilerWarnings. The Windows EXEC_BACKEND load-order
failure that affected v1 was fixed in v2 and hasn't recurred since.

Nothing's changed in the patches since the v2 posting on 2026-04-24, so
I haven't re-attached them.

A brief restatement of the ask, since the thread hasn't drawn review
yet: the series makes syncsw[] in sync.c extensible through a
register_sync_handler() API, so an extension can participate in the
checkpoint fsync pipeline instead of the handler table being fixed at
compile time. It's deliberately narrow, it doesn't change behavior for
any existing handler, and it doesn't depend on the broader extensible
storage manager work.

I'd welcome review, and I'm happy to rework the shape if registration
isn't the approach you'd want.

Greg Lamberson

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Masahiko Sawada 2026-07-30 21:48:39 Re: Fix "unexpected logical decoding status change" error; from concurrent logical decoding activation
Previous Message Alexander Korotkov 2026-07-30 21:19:03 Re: Function scan FDW pushdown