Hi, I'm not happy with compatiblity-definitions of char classes as its solved for example with [[:alnum:]] (when (featurep 'xemacs) (defcustom alnum "[a-zA-ZäöüßÄÖÜ0-9]" "Rexexp to specify the character class Change it, if you want to pick strings differently. XEmacs-users: `unibyte' and `multibyte' class is unused i.e. set to \".\"" :type 'regexp :group 'convenience)) It should be possible to detect the range from the buffer-coding-system instead customizing in manually. Any ideas how to perform this? Hi Steve, changing it at the bottom of things would take a lot of effort. Should solve some other tasks first, as beginning-of-defun-function, further enhancements to python-mode.el etc. Given there is no answer how to do it at lisp-level, please take `thing-at-point-utils.el', `thingatpt-utils-base.el' as is for the moment. It will provide a lot of useful functions IMHO. Cheers Andreas Röhler -- https://code.launchpad.net/s-x-emacs-werkstatt/ _______________________________________________ XEmacs-Beta mailing list XEmacs-Beta(a)xemacs.org http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta