latin encodings, mail-lib, and posix regexp classes

In an attempt to track down problems I'm having wrt to BBDB and Latin-9 characters, I've come across a problem. My understanding of these issues isn't very complete at the moment. My guess is that even when using the latin-unity package some of the characters are not marked as a word-constituant... For example: thisshouldbeßonewordright? Going to the start of that word, and hitting M-f advances to the ß. Actually, the ß in this mail is a Latin-1 character, but the effect is the same as with the iso-8859-15 encoded character. Basically, how can I get those extra characters in the syntax table in some kind of proper way? Or, are they supposed to be there already, and is my installation screwed in some way? Now onto what caused me to look for this. In mail-lib there are a couple things that concern me. First, the regexps used to match names won't work for names with latin encoded characters from iso-8859-15. For example, here is one regexp: (defconst mail-extr-first-letters (purecopy "A-Za-z")) Taking a look at the Emacs CVS, they use [:alpha:], which (in theory) should match the correct set of characters. But, I tried changing this, and it didn't work as expected -- does XEmacs support POSIX character classes? (string-match "[:alpha:]" "test") => nil I was going to go through the mail-extr.el file to find all the changes, but decided to hold up until I found out the deal with POSIX regexp classes. (since they're used fairly extensively in Emacs' mail-lib). Thanks for any help, -- Josh Huber