Stef> i'm using package-get to update my packages, and each time, Stef> i'm prompted to update the package list in a subdir of Stef> "/.xemacs", now i don't know how many of you have a .xemacs Stef> dir in the root dir but i suspect not many. in startup.el, Stef> it is set to : (defconst user-init-directory "/.xemacs/" Stef> "Directory where user-installed packages may go.") Stef> and i would therefore assume that code using this variable Stef> would prepend the missing '~' but not many packages use this Stef> variable in elisp, just where-was-i-db and package-get it Stef> seems. anyways, i can set it for myself, but i think that Stef> either it should be a default to put the '~' or code using Stef> it should prepend it ? after all it is very unlikely to Stef> have anything located in "/.xemacs/" right ? It looks like `user-init-directory', as Stef points out, is initialized to just "/.xemacs/", and is later used in combination with `init-file-user' to form the filename path in `load-user-init-file'. "package-get.el" just sets `package-get-user-package-location' to `user-init-directory'. This stuff needs to be gone over, I think. `user-init-directory' really ought to bear the `~' prefix; but the `init-file-user' thing has to keep working. I'm a little angry that the `init.el' and `.options.el' stuff got backed out. Why was that done? GNU Emacs and XEmacs often have incompatible configuration options. When you start one or the other using config files tainted by the other, it produces errors and an incomplete startup. It's not very difficult to run a grep and find ".emacs", then go there and edit the manuals and whatever else. If an "oreilly book says .emacs" then tough; the book's on paper or we'd email the author and it would be changed by the end of next week. I will submit patches for this if they will be accepted.