Charles> Adrian Aichner writes: Paul> In a CVS buffer, type 's' for cvs-status of a file. Attempt to load Paul> cvs-status library fails with: Cannot open load file: mule Charles> I've built the pcl-cvs package from sources and not had Charles> this problem (I run a mule-less XEmacs). Charles, I can confirm your finding after installing a locally built pcl-cvs 1.46, which I produced with (emacs-version) "XEmacs 21.1 (patch 10) \"Capitol Reef\" [Lucid] (i386-pc-win32) of Fri Jun 09 2000 on ZJ75T" and Local.rules containing BUILD_WITHOUT_MULE = t Charles> I suspect that the binary-kit packages must have somehow Charles> gotten infected with Mule; I think it has to do with how Charles> they were built, rather than the pcl-cvs code itself, Charles> which has no Mule dependencies. What are the rules for building packages? When should they be BUILD_WITHOUT_MULE = t ? Andreas, are you building them all BUILD_WITHOUT_MULE = ? Best regards, Adrian Charles> Oops, just after writing that last sentence I did a Charles> "grep mule pcl-cvs/*el" and came up with this: Charles> ;; chars sets. Ripped from cvstree Charles> (defvar cvstree-dstr-2byte-ready Charles> (when (featurep 'mule) Charles> (if (boundp 'current-language-environment) Charles> (string= current-language-environment "Japanese") Charles> t)) ; mule/emacs-19 Charles> in cvs-status.el... this is the only reference to Mule in the entire Charles> pcl-cvs package. Is this possibly the source of the problem?