The core problem that Emacs unexpectedly deletes file is still true, but the sequence for getting the error was incorrect.
I apparently opened the file with the default coding system. When I "describe-current-coding-system", I get Coding system for saving this buffer: * -- windows-1252-dos (alias: cp1252-dos) However, if I open the file with (setq save-buffer-coding-system 'windows-1252) ;; windows only then I don't get the error message buffer when I save. I remember now that I put in the "setq save-buffer-coding-system" long ago precisely to avoid the error message. To make this easier to check out, I've saved a small text file that generates the problem: http://GenerationalDynamics.com/gdgraphics/emacsproblem.txt Then if you want to have some fun, try the following: 1. Start Emacs. 2. Open emacsproblem.txt with default coding system. 3. Modify file and save it. 4. Get error message buffer, save as raw text. 5. Close emacs. Then repeat steps 1-5 over and over. Each time the file is saved, the file gets longer and longer, as the string of UTC characters gets longer and longer. I don't know, but my guess is that this is a feature not a bug. This character encoding stuff really drives me crazy. John ------------------- ***** Old message: 1. Open file, setting coding system as follows: (setq save-buffer-coding-system 'windows-1252) ;; windows only 2. Load a file containing \235 character. 3. Modify file and type Ctrl-X Ctrl-S. 4. Get error message buffer that begins: "These default coding systems were tried to encode text ..." 5. Get minibuffer message: "Select coding system (default raw-text): " 6. Decide you don't want to save, click "X" to close emacs. 7. Discover that the entire file has been deleted (though a recovery file was saved). John