It seems that by default the Postgresql server and frontend use the SQL_ASCII encoding. Given the Red Hat quest for Unicode "domination" :-), wouldn't it be better to switch defaults to Unicode? Besides, a utf-8 terminal and a psql using SQL_ASCII encoding do not get along very well :-(. Suggestions: - Switch initdb's default encoding to Unicode. - For psql, either: - have a /etc/skel/.psqlrc with "\encoding utf-8" (does not affect existing users :-() - add some magic to psql to determine whether the controlling tty has Unicode capability and set encoding accordingly (quick hack: check if LANG ends with ".UTF-8"?)
I will think about this. Thanks for the suggestions.
For the time being (to maintain regression test passes, among other reasons), I am going to keep the default encoding the way it is. Client encoding can be changed in dotfiles as you suggest. Databases can be initialized with other default encodings using `initdb --encoding=<encoding> -D <location of data directory>`.