Re: [18] Fix a few issues with the collation cache

Jeff Davis <pgsql@j-davis.com>

From: Jeff Davis <pgsql@j-davis.com>

To: pgsql-hackers@postgresql.org

Date: 2024-08-14T23:30:23Z

Lists: pgsql-hackers

Attachments

v2-0001-Minor-refactor-of-collation-cache.patch (text/x-patch) patch v2-0001
v2-0002-Tighten-up-make_libc_collator-and-make_icu_collat.patch (text/x-patch) patch v2-0002
v2-0003-Refactor-collation-version-check-into-new-functio.patch (text/x-patch) patch v2-0003
v2-0004-For-collation-cache-use-CollationCacheContext-for.patch (text/x-patch) patch v2-0004
v2-0005-Use-resource-owners-to-track-locale_t-and-ICU-col.patch (text/x-patch) patch v2-0005
v2-0006-Invalidate-collation-cache-when-appropriate.patch (text/x-patch) patch v2-0006

On Thu, 2024-08-08 at 12:24 -0700, Jeff Davis wrote:
> The collation cache, which maps collation oids to pg_locale_t
> objects,
> has a few longstanding issues:

Here's a patch set v2.

I changed it so that the pg_locale_t itself a resource kind, rather
than having separate locale_t and UCollator resource kinds. That
requires a bit more care to make sure that the pg_locale_t can be
initialized without leaking the locale_t or UCollator, but worked out
to be simpler overall.

A potential benefit of these changes is that, for eventual support of
multi-lib or an extension hook, improvements in the API here will make
things less error-prone.

Regards,
	Jeff Davis

Commits

Tighten up make_libc_collator() and make_icu_collator().
- ceeaaed87aa3 18.0 landed