Re: BUG #18691: Turkish Character Encoding

On Thu, Nov 7, 2024 at 1:08 AM PG Bug reporting form
<noreply@postgresql.org> wrote:
> The following bug has been logged on the website:
>
> Bug reference:      18691
> Logged by:          Ali Erkan
> Email address:      aerkan@gmail.com
> PostgreSQL version: 17.0
> Operating system:   Windows 10 & 11
> Description:
>
> With the latest Windows update, Microsoft has changed Turkish_Turkey.1254 to
> Turkish_Türkiye.1254.
> PostgreSQL installation gets an error during database and service creation,
> and the database and Windows server are not created.
> It seems to be related to the name changes in the locale.

Hello,

PostgreSQL can't work with locale names that contain non-ASCII
characters.  In 17.0 and 16.4, the Windows system library crashes
while we are initialising a new database cluster.  For 17.1 and 16.5
(due out soon) we've recently made a change so that it still won't
work, but it'll display and error message to explain why:

https://github.com/postgres/postgres/commit/9c7acc33307bf971ff5fe4b7f7431b7aa93052ac

As hinted at the bottom of that commit message, others have reported
success by using Microsoft Locale Builder to create a new locale with
an acceptable name (that is, without the problematic ü character), and
apparently there are some good articles on the Turkish language web
about how to do that.  If you find a good one let us know so we can
share it with the next person.  It's also possible to use "tr-TR"
instead but I'm not sure if the EDB graphical tooling currently has a
nice way to do that, if that's what you're using.