Hello Fellow Monks,

I'm have a load of unicode/locale problems (more posts may be forthcoming) on a properly configured Linux system and I'm hoping someone can point out some of the reasons things are not turning out how I want them to turn out.

Observe:

# note: system locale works properly
$ LC_CTYPE=fi_FI.UTF-8; export LC_CTYPE

# WITH use locale:
$ perl -CSA -e 'use utf8; use open ":std", IO => ":utf8"; use locale; 
+use POSIX qw(locale_h); print "LC_CTYPE: ", setlocale(LC_CTYPE), "\n"
+; $str = "säv sov san"; my @arr = $str =~ /\b(\w{3})\b/g; print "", j
+oin("|", @arr), "\n";'
LC_CTYPE: fi_FI.UTF-8
sov|san

# WITHOUT use locale:
$ perl -CSA -e 'use utf8; use open ":std", IO => ":utf8"; use POSIX qw
+(locale_h); print "LC_CTYPE: ", setlocale(LC_CTYPE), "\n"; $str = "sä
+v sov san"; my @arr = $str =~ /\b(\w{3})\b/g; print "", join("|", @ar
+r), "\n";'
LC_CTYPE: fi_FI.UTF-8
säv|sov|san
[download]

... use locale breaks '\w'. It doesn't consider "säv" to be consisting of "word" characters. Since it's a Nordic UTF-8 locale, '\w' should contain "öäå". It works without it, strangely enough.

BTW, even when dropped the -CSA unicode argument to Perl, use open works as expected with both :utf8 and :locale, so Perl does read my locale settings.

What gives?

PS: Perl v5.10.1 on Debian "unstable".