Content-Type: text/plain; charset="iso-8859-15"
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
Subject: Re: ICU for global collation
From: "Daniel Verite" <daniel@manitou-mail.org>
To: "Julien Rouhaud" <rjuju123@gmail.com>
Cc: Peter Eisentraut
 <peter.eisentraut@enterprisedb.com>,
 Pgsql-Hackers <pgsql-hackers@postgresql.org>
In-Reply-To: <20220107090329.lnsy6tnodrrbmyy5@jrouhaud>
Date: Fri, 07 Jan 2022 18:19:09 +0100
Message-Id: <33af6926-cc9f-486d-8db1-41c575b9a9cb@manitou-mail.org>
Archived-At: <https://www.postgresql.org/message-id/33af6926-cc9f-486d-8db1-41c575b9a9cb%40manitou-mail.org>
Precedence: bulk

	Julien Rouhaud wrote:

> If you want a database with an ICU default collation the lc_collate
> and lc_ctype should inherit what's in the template database and not
> what was provided in the LOCALE I think.  You could still probably
> overload them in some scenario, but without a list of what isn't
> ICU-aware I can't really be sure of how often one might have to do
> it.

I guess we'd need that when creating a database with a different
encoding than the template databases, at least.

About what's not ICU-aware, I believe the most significant part in
core is the Full Text Search parser.
It doesn't care about sorting strings, but it relies on the functions
from POSIX <ctype.h>, which depend on LC_CTYPE
(it looks however that this could be improved by following
what has been done in backend/regex/regc_pg_locale.c, which has
comparable needs and calls ICU functions when applicable).

Also, any extension is potentially concerned. Surely many
extensions call functions from ctype.h assuming that
the current value of LC_CTYPE works with the data they handle.


Best regards,
--=20
Daniel V=E9rit=E9
PostgreSQL-powered mailer: https://www.manitou-mail.org
Twitter: @DanielVerite