Evergreen cannot sort the strings according to selected locale

[Eva Cerninakova]

For the sorting of any list, only the Unicode sorting is used in Evergreen. This means that, e.g. for the Czech language, the sorting does not correspond to the order of letters in the Czech alphabet.
This applies both to any kind of OPAC result (Search result screen, Browsing, Shelf Browsing) and Staff client "list" sorting (rows in various types of grids, bucket names etc.).

[Eva Cerninakova (ece)]

Still issue in current main branch

[Jason Stephenson (jstephenson)]

Most of the sorting is done in the database, so the problem is how we create the database.

We should provide options in the database setup script for the installer to specify the database language settings when it is created.

You can change the collation and encoding settings of your current database by doing a pg_dump from the original and a pg_restore into a new database with the desired settings. This will be a lengthy process on a moderate to large-sized database. Evergreen will have to down for the duration of the process, or you will lose any transactions added during that time.

There really is not a viable way to fix this with code, not even taking into account that individual collations can be set per table column.

For more information, I refer you to the PostgreSQL Localization documentation: https://www.postgresql.org/docs/15/charset.html