Julien Rouhaud wrote:
> > I guess there's still the possibility of requiring that the ICU db-wide
> > collation of the new database does exist in the template database,
> > and then the CREATE DATABASE would refer to that collation instead of
> > an independent locale string.
>
> That could work. However if having the collation in the template database a
> strict requirement the something should also be done for initdb, and it will
> probably be a bigger problem.
If CREATE DATABASE referred to a collation in the template db,
either that collation already exists, or the user would have to add it
to the template db with CREATE COLLATION.
initdb already populates the template databases with a full set of
ICU collations through pg_import_system_collations().
I don't quite see what change you're seeing that would be needed in
initdb.
Best regards,
--
Daniel Vérité
PostgreSQL-powered mailer: https://www.manitou-mail.org
Twitter: @DanielVerite