Solved PostgreSQL 10.6 and ICU collations?

Hi. Now that PG 10.x has stopped using its ICU hack (as it did up to versions 9.6.x), how do I make the collations usable? I have these packages installed:
postgresql10-client-10.6_2 PostgreSQL database (client)
postgresql10-contrib-10.6_1 The contrib utilities from the PostgreSQL distribution
postgresql10-server-10.6_2 PostgreSQL is the most advanced open-source database available anywhere
icu-63.1_1,1 International Components for Unicode (from IBM)


But even simplest collations don't work.
Code:
postgres=# \l
          Name          |  Owner   | Encoding |   Collate   |    Ctype    |   Access privileges   
------------------------+----------+----------+-------------+-------------+-----------------------
foo                     | binaaz   | UTF8     | en_US.UTF-8 | en_US.UTF-8 | 
postgres=# \c foo
You are now connected to database "foo" as user "postgres".
foo=# CREATE TABLE foo (foo text COLLATE "en-x-icu");
ERROR:  collation "en-x-icu" for encoding "UTF8" does not exist
LINE 1: CREATE TABLE foo (foo text COLLATE "en-x-icu");
 
Woops, I can see that in a newly installed PG 10.6 and cluster initialized by initdb all these collations are available in pg_collation schema. Is there any good way to install them for an existing cluster?
 
Back
Top