Arch Linux

Description:
On most distributions, PostgreSQL defaults to UTF-8 encoding, as they have the LOCALE or LANG environment variable set with an UTF-8 locale. Most users expect recent database engines to "just work" with Unicode data without any special configuration.

On Arch Linux, however, even if the user has LOCALE=en_US.UTF-8 set in their rc.conf, PostgreSQL gets initialized with a C locale because it relies on LC_COLLATE. Arch's default /etc/profile overrides LC_COLLATE=C

Personally I think this is a good override, but it just doesn't play nice with PostgreSQL. Thus I suggest changing initdb in /etc/rc.d/postgresql to use the user's preferred LOCALE:
- su - postgres -c "/usr/bin/initdb -D $PGROOT/data"
+ su - postgres -c "/usr/bin/initdb -D $PGROOT/data --locale $LOCALE"

(See also: related bug  FS#12432  to specify initdb options from /etc/conf.d)



Steps to reproduce:
# grep ^LOCALE /etc/rc.conf
LOCALE="en_US.UTF-8"
# grep ^LC_COLLATE /etc/profile
LC_COLLATE="C"
# pacman -S postgresql
...
# /etc/rc.d/postgresql start
:: Starting PostgreSQL [BUSY]
...
The database cluster will be initialized with locale C.
The default database encoding has accordingly been set to SQL_ASCII.
...