Moritz Schepp <moritz(dot)schepp(at)gmail(dot)com> writes:
> My problem is that I don't (even) get the german collation to work, the
> following query yields 'f':
> SELECT 'a' COLLATE "german" LIKE 'ä' COLLATE "german";
> Shouldn't that be working?
No. Postgres interprets COLLATE as having to do only with sort ordering.
What you want here, IIUC, is to do accent-stripping on the data and/or the
LIKE pattern before you compare them. Take a look at the contrib/unaccent
module for one solution.
regards, tom lane