Steve Youngs writes: I still don't really like this. AFAICS this makes testing hard. Having `byte-compile-and-load-file' error because you're using 21.5 doesn't sound like a great idea to me. Sure you can work around it with fmakunbound, but that's kind of ugly. In general, somebody who `load's rather than `require's bytedecl probably has a good reason to do so. Is there anything wrong with putting `(provide 'bytedecl)' in lisp/bytecomp-runtime.el? For example, maybe SXEmacs has diverged, in which case you may not want bytedecl overwriting your macros in older versions without the provide. (I'm going to be out of town for a few days, so don't expect further discussion from me until I get back.) _______________________________________________ XEmacs-Beta mailing list XEmacs-Beta(a)xemacs.org http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta